J.Adamczewski-Musch,
M.Al-Turany, D.Bertini, H.G.Essel, S.Linev
23 October 2008
Content
The
Go4 Analysis Framework Introduction V3.4
2.1 New
features in Go4 v3.04
(October08)
2.2 New
features in Go4 v3.03 (May07)
2.3 New features in Go4 v3.02 (July06)
2.4 New features in Go4 v3.01 (May06)
2.5 New features in Go4 v3.00
(November05)
2.6 New features in Go4 v2.10 (June05)
2.7 New features in Go4 v2.9
(February05)
2.8 New features in Go4 v2.8
(September04)
2.9 New features in Go4 v2.7 (June04)
2.10 New features in Go4 v2.6 (May04)
2.11 New features in Go4 v2.5
(December03)
2.12 New features in Go4 v2.4 (August03)
2.13 New features in Go4 v2.3 (May03)
2.14 New features in Go4 v2.2 (April03)
3.1.1 Go4 tasks with all communications
3.1.3 Other analysis functions
4.2 Event classes, interface to MBS
4.5.1 Batch or command line mode
4.5.2 Client mode controlled by Go4 GUI
4.5.3 Analysis in server mode for
multiple Go4 GUIs
4.5.4 MainUserAnalysis example
4.5.8 Start-up of the analysis slave
4.5.9 Submit settings and run analysis
4.5.10 Shutdown of the analysis client
4.5.11 Disconnect or shutdown analysis
server
5.2.3 Using the GUI with rsh or ssh
5.3 Simple example with one step
5.3.1 Main program and analysis
5.3.5 Auto-save file mechanism
5.4.1 Main program and analysis
5.4.4 Auto-save file mechanism
5.5.1 Main program and analysis:
6.1.1 File, Tools, Analysis menus
6.3.1 Launch analysis task in client
mode
6.3.2 Launch analysis task in server
mode
6.3.3 Connect to existing analysis
server
6.4.2 Analysis terminal window
6.4.3 Macro execution in the analysis
6.4.4 Auto-save file mechanism
6.4.6 User defined event sources
6.5.3 Analysis folder controls
6.5.7 Histogram server connection
6.5.8 Resetting and deleting objects
6.6.2 Remote mode (dynamic list
histogram)
6.6.3 Creating a new histogram
6.7.6 Channel and window markers
6.8.1 Conditions editing in viewpanel
marker editor
6.8.4 Conditions bound to pictures
6.11.3 Parameters containing fitters
6.13 Histogram/condition information
7 Analysis Server for ROOT macros
7.1 Methods for object registration
7.2 Methods for run control and
execution
8 Control of remote Go4 analysis from
a ROOT session
8.3 Controlling the analysis by command
9 The Go4 Composite Event Classes
11 Table of Menu Keyboard Shortcuts
Layout used in this document:
Text Times New Roman, 10 pt
Verbatim text Courier new 10 pt
Menu items Arial bold 9 pt
Class names Arial italics , 9 pt
Methods () Arial italics , 9 pt
Go4 screenshots Style
Window, Font Arial 11pt
Einfügen->Referenz->Querverweis:
Überschrift+Überschriftnummer/Seitenzahl
Einfügen->Referenz->Index
und Verzeichnisse: Eintrag festlegen, Indexeintrag+Aktuelle Seite. (search for Feld)
Index entries can be edited in text (first:second)
Index
aktualisieren (RMB)
Inhaltsverzeichnis
aktualisieren (RMB)
a.
Mbs API: streamserver connection timeout was not working correctly (leads to hangup of analysis
control when no data is delivered from streamserver).
b.
Mbs API: several errors at reading of *.lmd files with new event format 100,1
(DABC)
c.
MbsAPI , for f_stccomm.c file. Fixes problem with connecting 64 bit machine to MBS events/stream/trasnport
server. False usage of select()
function.
d.
TreeViewer swapped x/y/z coordinates, convention is
TTree::Draw("z:y:x"))
e.
Viewpanel: "Produce Picture" did not save all draw
options to picture
f.
Viewpanel: Mismatch between Go4 viewpanel range (full visible range) and ROOT user
range (referring to low edges of bins) could cause slight shift of x axis range
on canvas refresh
g.
TGo4Browser: Arrays fVisibleColumns and fIndexes has 1 item less
than required
h.
QRootApplication: in constructor numc argument must be delivered as reference.
i.
TGo4MBSViewer: status record must be cleared in constructor.
j.
Fit package concerning parameters handling when some parameters are fixed
a.
Modifications in makefiles - now only in one place in
Makefile.config one should specify platform - Linux, Solaris or Win32. Other
small changes in makefiles
b.
Small
adjustments for the new ROOT (5.17.05) browser.
c.
Adjustments
for modifications in ROOT signal-slot mechanism syntax (ROOT >=5.19/02);
this caused viewpanel crash.
d.
Two
ROOT libraries (libTree and libGpad) added to Go4 rootmap file that user Go4
analysis library can be loaded in CINT session.
e.
ThreadManager
workaround for ROOT bug in TThread::Delete() (ROOT bug report 31085): for some compilers, Go4 GUI crashed when shutting down or disconnecting analysis.
1.
Viewpanel
a.
Marker
editor: A point- or region marker and its label will pop to the pad foreground when it is selected
with left mouse button. Additionally, selection of a marker in the combo box of
the editor will let it appear frontmost.
b.
In superimpose mode selected histogram can be moved on the top of complete histogram stack via new
menu command "Select/show histo on top".
c.
Draw
options enhanced: support for TGraph draw modes and TGraphErrors error style. Reorganization of draw options for
TH1/TH2. New draw options tool for line, marker, and fill colours of histograms
and graphs.
d.
Menu
"Select" to chose active object from superimposed histograms
and graphs.
e.
Autoscale
checkbox as shortcut on top of each viewpanel
f.
Improvement in
speed of view panel redraw (up to factor of 2).
2.
Fitpanel
improvement: keep y-scaling when fitting on x subrange of histogram
3.
New
Zoom toolbar: added buttons for scaling z-axis of 2d histograms.
4.
New
icons for zoom toolbar and draw options toolbar.
5.
New
additional draw options toolbar to select commonly used drawing options by
buttons (lin/log, line, histo, some 2d styles). The new toolbar is displayed
via the RMB options pull down menu.
6.
New
example macro scalex.C to scale x-axis of histogram with linear calibration
function
7.
Settings
menu: "Show event status" selectable as default
pad option.
Settings menu: "Statistics Box..." dialog to define default pad
options for histogram statistics.
8.
TGo4Picture: new
method AddSpecialObject() to
add any ROOT graphical object (text labels, markers) to the picture
9.
Improvement in TGo4MbsFile for
partial read of lmd file: Corrected mismatch between first event index and real
event number (before: index=event number-1).
10.
TGo4MbsFile: now
can also read list-mode data of old event formats type 4,1 and 4,2. Event will
be converted implicitly into format 10,1 for further processing: User unpack
processor gets TGo4MbsEvent with
one TGo4MbsSubevent that
contains all event data.
11.
GUI command
interface TGo4AbstractInterface.
Added methods:
- GetViewPanelName() - returns view panel name
- SetViewPanelName() - changes view panel name
- RedrawPanel() - updates view panel view
- RedrawItem() - updates all views of specified items
- FindViewPanel() - searches for view panel of specified name
- GetActiveViewPanel() - returns currently active view panel
12.
Maintenance:
a.
Some Makefile and build skript improvements
b.
Added missing includes for <math.h>, required by
some compilers
c.
Due to changes in ROOT in many Go4 files includes like
TROOT.h, TMath.h, TList.h are
missing. Sometimes user should also include these files in user code.
d.
In latest ROOT TBuffer class becomes abstract, therefore one cannot use it directly
in the code. Instead, TBufferFile class
must be used.
e.
Adjustment of Makefile
because of changes in default libraries for ROOT >= 5.13/04 (separated libSpectrum.so)
f.
Adjusted Go4ThreadManager
package due to changes in TTimer copy
constructor for ROOT versions > 5.12.00
g.
Some bug fixes concerning compilation against old ROOT
versions 4.08
13.
Bug
fix
a.
for changes in ROOT>v5.14 pad cleanup: Viewpanel with go4 markers on subpads crashed
when closed or cleared.
b.
1-d histogram drawing. Due to some features of ROOT
histogram painter several draw options (lin, barchart and others) not working
after TH1::SetSumw2() is
called - in there Sumw2 array sum of squares of weights is accumulated.
Modification in Go4 code were done to avoid Sumw2 arrays when it not necessary.
c.
in Go4Socket library (missing include) because of
changes in ROOT version 5.14-00
d.
Problems with view panel scaling functionality when
build with gcc4.0.x compiler (FC5); fixed.
1.
Analsis framework: TGo4EventElement now implements default
method Fill() that calls virtual function TGo4EventSource::BuildEvent(). As a consequence, for a simple analysis the user
only has to implement BuildEvent()
method in his processor class. There is no need to
develop a user output event class. Even if a user output event class shall be used,
methods Fill() and Init() are not necessarily needed for a standard analysis. Go4ExampleSimple and Go4Example1Step were changed accordingly.
2.
Analysis framework: TGo4EventProcessor now implements BuildEvent() and can be used in
steps which are only used as handle for event input (branched steps).
3.
Macro usage: Analysis defines __
4.
Parameter editor offers popup menu GetFromFitPanel for embedded
fitters to update fitter settings from the current fit editor. Useful for
calibration parameters that should be fitted interactively to spectra (see Go4Example2Step).
5.
Rebin in GUI. Now when histogram will be rebinned via right-mouse
menu or via ROOT graphical editor, rebinning will be kept when histogram will
be updated next time from analysis. Many views of the same histogram with
different binning are possible. Binning also kept in hot-start file. TGo4Picture has new SetRebinX(), SetRebinY() methods to configure rebinning of displayed
histogram.
6.
All Go4 macros put into new subfolder $
New macros: savecond.C and saveparam.C to create macros to set conditions and parameters to
their current values (see 4.5.6, page 26).
7.
Bugfixes:
a.
Access to RFIO root
files from Go4 GUI browser was not possible (at GSI), since internal functions
of libRFIO.so
were shadowed by functions of GSI event lib with same names. Solved by separating
Go4 event library package into different modules for analysis and GUI task.
b.
Analysis server
executed UserPostLoop() each time a GUI client was disconnected. Disabled.
c.
Several changes
concerning the cleanup mechanism in GUI object manager
d.
AnalysisClient in CINT mode showed thread deadlock for ROOT versions>
5.02-00
e.
Start client dialog
selects correct analysis directory when choosing the analysis executable
1.
New script
command line widget for GUI: Allows execution of ROOT commands or macros
within Go4 GUI task. Moreover, Go4 hotstart scripts may be invoked here at any
time. The widget offers a file dialog to search for *.C and *.hotstart files.
It also has a selector dialog of preloaded commodity functions for histogram manipulation
(rebinning, addition, projection, etc.). These function template calls may be
completed with existing histogram names by dragging histogram items from the
browser and dropping them on the empty command argument. The history of the
command line may be saved to the current Go4 settings file .go4/go4localrc and
is then restored on next startup. (See 6.17, page 81).
2. New GUI command interface class TGo4AbstractInterface. It
can be accessed by handle "go4->" in GUI
command line. This makes it possible to interact with Go4 GUI views and browser
objects in a ROOT/Go4 script. Additionally, all remote analysis control
commands are available here, like in the hot start scripts. Method reference of
TGo4AbstractInterface is
available in the Go4 help viewer (type "help" in GUI command line, or
use Help►GUI
commandline menu of Go4 main window). Example scripts using this interface are at $
3.
New general
marker label settings dialog. In main window menu Settings►Panel
Defaults►Marker labels.., a
checkbox dialog offers to switch all label properties of the region and point markers
(visibility and information displayed in the label). These settings have effect
on all new markers of the view panel marker editor. They are saved in the go4
preferences file .go4/go4localrc. (see 6.7.6, page 64)
4.
Plain
ROOT canvases in files are better displayed.
5.
New settings feature Settings►Preferences►Fetch
when saving. If enabled, the save browser / save memory button of the file toolbar will
refresh all browser item objects from analysis before saving. Thus the ROOT
file will contain a snapshot of all analysis objects. Otherwise, only the
already fetched objects are saved.
6.
Zoom
tools "set scale" dialog upgraded to non modal MDI
widget. This will appear always on top of workspace widgets and refers to
currently selected view panel pad. Changes include some bug fixes concerning
the range settings of 2d histograms, and the auto-scale property.
7.
MBS
monitor tool: If monitoring switched on, calculation of rates is now
done in Go4, averaged over update time. Parameters of MBS monitor are stored in
Go4 settings file.
8.
TGo4Interface: new
method ExecuteLine to
remotely do CINT call from Go4 master process in the remote slave process
9.
View panel
superimpose mode improvements:
a.
is not changed anymore after superimposed draw of
FitPanel results, i.e. fitter data histogram can now be replaced just by drag and
drop on the view panel
b.
existing axis labels of first histogram are kept
10.
FitPanel
settings are saved/restored in go4 settings file
11.
Fit
GUI: Enhanced draw styles for TGraph
12.
Bugfixes:
a.
Workaround for ROOT crash in histogram rebin editor: Selecting
a histogram in view panel for rebin with the ROOT attributes editor leads to
segmentation violation when original histogram was replaced or deleted.
b.
Crash
in Go4 markers/conditions when histogram in view panel was replaced by drag and
drop.
c.
Update of histogram in GUI failed when histogram
dimensions (ranges) were changed in analysis
d.
Position and size of histogram statistic label may now
be saved in Go4 picture objects. Thus these properties can be restored on Go4
hot start.
e.
Crash
on closing last non-minimized window in view panel
f.
Problem
with empty TGraph as data source
in Fitter
g.
Crash
when FitPanel histogram under work was replaced or deleted in view panel. FitPanel did not react automatically
on changes, happening with histograms (or graphs), displayed on view panel.
Therefore, when superimpose mode was switched off, fitted histogram disappeared
from view panel (and also deleted), while fitter still has pointer on that histogram.
Now FitPanel slot in object
manager registered also against all histogram, used in fitting. If histogram is
deleted, FitPanel will be
automatically refreshed.
h.
Histogram title could not be switched off in
superimpose mode in view panel
13.
Improvements in make files
14.
Adjustments of includes due to changes in new ROOT
version 5.10
1.
Marker editor in view
panel allows for marking channels or
windows. Labels and arrows can be created. All marker elements can be saved and
restored.
2.
New ROOT graphical editor can be called
from view panel. The
editor dynamically adjusts to the graphical object selected by LMB.
3.
View panel window title: can
optionally be set by user and may be kept constant. If a TGo4Picture is displayed, the
picture name defines the view panel title.
4.
Condition editor: the cursor mode has
been removed because the functionality is now provided by the markers
5.
Condition, markers and labels: Implemented correct ROOT streamer (bug fix), i.e. saving and
loading these objects to and from ROOT files is possible with fully recovered
functionality and graphical properties. Support of pad display in linear and log
scale (bug fix). Additional controls in RMB menu of ROOT (set ranges, location,
save default properties, reset). Default label setup stored with Go4 GUI settings.
6.
Polygon condition: Implemented statistics
functions for work histogram under the cut (integral, mean, rms, etc.). Enabled
InsertPoint and RemovePoint functions in RMB menu (bug fix).
7.
Fit GUI: Selection
between sigma and FWHM (default) by Settings►Recalculate gauss width.
Fit results may be printed to terminal or Go4 log file
output.
8.
1D drawing: ROOT "L" (line) "C"
(curve) "B" (bar chart) "P0" (poly-marker) line styles supported.
9.
Histograms: re-binning, projections, and
profiles supported (standard ROOT methods with RMB). Automatic “synchronize
with memory” on pad click to get newly created histograms.
10.
Histogram client: monitoring implemented
(auto-update). Drag and drop support. Display error message when server
connection is not available (bug fix). Store server specification in Go4
settings.
11.
File store: Storing objects into a ROOT
file a title is prompted. This title can be seen in the Go4 browser and the
ROOT browser.
12.
UserObjects folder: With AddObject(...) histograms, parameters and conditions can be put into folders of
the
UserObjects folder. They can be located there by the standard Get methods, e.g. GetHistogram(). Editors work also with objects in these folders. Note: object names must be unique!
13.
Log window: Empty messages are now
suppressed (bug fix).
14.
QtRoot interface: bug fix concerning
initialization order of X11 system (ROOT init now before Qt init). Lead to
crash of the main GUI on newer Linux systems when using Qt versions > 3.1
(FEDORA2, SuSe9.1)
15.
Thread manager:
bug fix: adjusted default exception handling to
work with newer libpthread.so that uses one process for all threads (e.g. FEDORA2). This lead to
a crash when Go4 threads were canceled (shutdown of the go4 GUI).
16.
Analysis Framework:
bug fix: analysis without analysis step (UserEventFunc() only) again possible.
17.
Client startup script: full PATH and LD_LIBRARY_PATH of the Go4 GUI environment
is passed to the analysis process.
1.
Keyboard shortcuts (Alt-1 to Alt-5) to select browser tabs (File, Monitor, Remote, Memory, Histogram
client). Items are selectable with arrow keys (left-right to unfold and shrink
subfolders). Return key acts as double click.
2.
MBS event classes improvements: Method TGo4MbsSubEvent::IsFilled() checks if the sub-event was filled in the previous event built. Iterator
TGo4MbsEvent::NextSubEvent() by default delivers newly filled sub-events only, suppressing
existing sub-events in list of non used ids.
Sub-event data field re-uses the memory allocated by libgsievent instead of copying it to own buffers. New method TGo4MbsEvent::SetPrintEvent() to set verbose mode for the next n events. Format changes in TGo4MbsEvent::PrintEvent().
3.
Performance improvements of analysis
framework in step manager, dynamic list and MBS event classes.
4.
New EventInfo toolwindow to control
printout of an event sample in remote or local terminal. Optionally the user
implemented PrintEvent() method, or the ROOT TTree::Show() output may be used. May control the arguments of TGo4MbsEvent::SetPrintEvent(). Supports drag and drop for event names from remote browser.
5.
Display total memory consumption of
histograms and conditions at the end of PrintHistograms() and PrintConditions() execution, respectively.
6.
TCanvas support in file browser improved: Histograms saved inside a TCanvas in a ROOT file will appear in memory browser whenever this canvas
is displayed
7.
Analysis Terminal window: Limitation of
text history buffer to 100 Kb by default, may be changed in settings menu.
Disabled text wrapping in output for scrollbars.
8.
Scale values dialog window extended by
zmin and zmax fields. Allows setting minimum and maximum thresholds for channel
contents of 2d histograms when auto scale is off.
9.
Conservation of TLateX textfields when changing draw
style or histogram statistics boxes visibility
10.
File browser open file dialog
allows multiple file selection
11.
Analysis configuration window: remember
path to previous selected file in event source, auto-save, and preferences dialogs.
Some layout cleanups.
12.
Superimpose of histograms with same name
from different files possible if overwrite mode is deselected in memory
browser. Histograms will be copied to memory browser with cycle numbers added
to names.
13.
Bugfix: Superimpose THStack does not crash anymore
when deleting histograms
14.
Bugfix: Crash after closing and
re-opening view panel for same histogram with different sub-pad divisions
15.
Bugfix: Analysis did stop when an
analysis step without event processor is disabled
16.
Bugfix: histogram bound to condition was
not fetched from analysis when double clicking on remote condition icon
17.
Bugfix: Double click on histogram in
divided view panel did pop up this histogram magnified in a new view panel, but
did not initialize view panel colours and crosshair settings correctly.
1.
New Go4 Hotstart: The current setup of the GUI (analysis name and settings, view
panel geometry, objects in memory
and monitor browser, displayed objects in pads) may be saved to a hot start
script file (postfix ".hotstart") from the Settings►Generate hotstart menu. The script name may be passed as argument on next Go4 GUI
startup (e.g. "go4 mysetup"),
which will launch the analysis and restore the settings (e.g. from file "mysetup.hotstart").
2.
New TGo4ExportManager class
transforms and saves ROOT objects into other formats. Currently supported:
plain ASCII (*.hdat, *.gdat) and Radware/gf3 (*.spe). An export filter is
available in the GUI memory browser to save selected objects.
3.
Redesign of Go4 Auto-save mechanism. Subfolders are
mapped as TDirectory in TFile
now, thus improving performance for large number of objects. Auto-save file is
closed after each write, avoiding invalid file states in case of analysis
crash. Dynamic list entries are saved as independent objects.
4.
Example macro Go4Example2Step/convertfile.C converts all histograms and graphs from ROOT file into ASCII files,
conserving the subfolder hierarchy.
5.
New TGo4StepFactory class can be
used as standard step factory to simplify the setup of analysis steps for small
analyses. New example package Go4Example1Step shows the usage.
6.
The TGo4Analysis class can now
be used as standard analysis class. New
example package Go4ExampleSimple shows the usage.
7.
New view panel has size of previously active view panel. Default view panel
starting size is stored in settings and recovered on next Go4 startup.
8.
View panel: Switch on/off histogram title display in options menu.
9.
View panel: Switch on/off crosshair for each pad in options menu. Default
crosshair mode can be selected in main window settings menu and is saved and restored
by Go4 settings. Crosshair mode button in condition editor has been removed.
10.
View panel: Default background color can be selected in main window settings
menu and is saved/restored by Go4 settings.
11.
TCanvas objects in analysis task may be send and displayed on GUI. Works
both for memory and monitoring list.
12.
Support of TMultiGraph objects in
analysis and GUI (display, memory and monitoring list update).
13.
New draw option TASImage for 2 dim
histograms in Go4GUI. May improve rendering speed for large maps when updating
and resizing the canvas. Offers own palette editor in right mouse button popup
menu.
14.
Parameter editor: Added column to display the source code comments for each parameter
class member as description.
15.
Condition editor: General editor has button to create a new condition. New condition
is defined in a dialog window and is put into general editor. May be sent to
analysis for registration, or saved into a file then. All types of new
conditions (window, polygon, array of these with variable size) are supported.
16.
Object editors (condition, parameter, dynamic list) may save and load objects
from/to ROOT files.
17.
Status messages of object editors appear in bottom status line of Go4 main window.
18.
Support of dynamic list
entries in file browse: Editor opens on double click.
19.
Histogram and Condition info windows: Object size now takes into account real data size on heap.
20.
New analysis toolbar button for
"re-submit and start" shortcut. Useful when file shall be
re-read from the beginning after changing something in the setup.
21.
Auto-save may be disabled completely from analysis configuration GUI.
22.
New mode for TGo4MbsFile (*.lmd)
wildcard/metafile input: Auto-save file may change its name whenever input file
is changed. Name is automatically derived from input filename. Old behavior
(one auto-save summing up all inputs) is still possible. This can be switched
with method TGo4Analysis::SetAutoSaveFileChange(bool
).
23.
End of .lmd file input gives informational message instead of error message.
24.
Bug fix: avoid log-file
crash when Go4 is started in directory without
write access.
25.
Bug fix in Go4 Mainwindow
exit dialog. Exit via window "x" icon
works properly now, too.
26.
Some adjustments to work with
ROOT versions > 4.00 in Go4Fit and qtroot packages
1.
Histograms may be bound to
conditions by method TGo4Conditions::SetHistogram(). The bound histogram will be fetched automatically in GUI whenever
condition is edited.
2.
TGo4Picture can contain
conditions together with histogram objects.
3.
General condition editor in
addition to the condition specific editors. Supports drag and drop of condition
icons and conditions linked to TGo4Pictures.
4.
Warning label for unsaved changes
in condition editor, and in dynamic list editor.
5.
Condition editor cursor tab can
make copies of the current cursor marker. For printouts with multiple markers.
6.
Analysis log window in GUI
displays date and time of last refresh.
7.
New histogram status window,
and condition status window in GUI.
8.
Redesign of GUI object
management: Added drag and drop support of TGraph, TGo4Picture from all browsers. Bug fix and improvements in
histogram superimpose mode.
9.
Monitoring list supports TGraph, TGo4Picture, and THStack.
10.
Logfile mechanism for GUI
actions. Log output configurable in Settings menu. Logging output on demand
from condition editor, histogram and condition status windows.
11.
View pane can turn on or off
histogram statistics box.
12.
View panel supports fix/auto scale
modes for TH1, THStack, and TGraph objects.
13.
View panel resize speed
improved (redraw only at the end of resize action). View panel does not start
in full screen mode anymore.
14.
Analysis terminal: New buttons
for clearing the terminal, PrintHistograms, PrintConditions. Command line has
shortcut “@” for “TGo4Analysis::Instance()->”. “KillAnalysis” button
buffered with confirmation dialog window.
15.
“Quit Go4” button buffered with
confirmation dialog window.
16.
Dynamic list editor can change
the global dynamic list interval for analysis.
17.
Reorganization of GUI icons.
18.
Performance improvements in
TTimers of Go4 kernel: Removed Turn On/Off statements.
19.
New method TGo4Analysis::NextMatchingObject() for search in analysis objects with wildcard expression.
20.
Analysis: PrintHistograms(), PrintConditions() supports wildcard expressions for output list selection.
21.
New methods: TGo4Analysis::StoreParameter,
StoreCondition, StoreFitter, StoreFolder to write
these objects into event store of an analysis step. Event number will be
appended to object keys for parameter logging.
22.
Consistency checks of analysis
steps can be disabled by new method TGo4Analysis::SetStepChecking(bool). For setting up of non serial type analysis steps with own user
management.
23. TGo4MbsEvent::PrintEvent() extended to display headers and also data field contents of sub-events.
24.
New methods: TGo4MbsEvent::GetMbsBufferHeader(),
TGo4MbsSource::GetBufferHeader() to access the
buffer headers of list-mode files. Implemented example in Go4Example2Step.
25.
Go4 GSI histogram server also
exports TGraph objects as histograms (if possible).
26.
Implementation of TGo4Condition::Paint() to display Go4 conditions in regular ROOT environment. Conditions
may be drawn on TPad which already contains a histogram. New classes for condition painters
and condition views.
27.
Reorganization of the
distribution make files.
1.
New Package Go4Log to handle
all messages and log file. This replaces the old package Go4Trace. Static
method TGo4Log::Message(char*,
...) can be called everywhere to display text on
terminal and optionally write to log file. Modified Go4 message prompt.
2.
Header information of MBS list-mode
data files accessible by new methods s_filhe* TGo4MbsSource::GetInfoHeader()
and s_filhe* TGo4MbsEvent::GetMbsSourceHeader().
3.
Event source class TGo4MbsRandom to deliver random spectra into MBS events without connection to MBS
node or reading list-mode file. Matches event structure of standard example Go4Example2Step.
4.
TGo4Picture objects can be used in the monitoring list.
5.
Changes in Analysis
configuration window: Number of events, start/stop/skip events may be
specified; tag file name and optional socket timeout. File browser for event
source files. Auto-save interval now refers to time (seconds) instead number of
events. Modified layout.
6.
Dynamic list editor with button
to PrintAll
dynamic list entries on analysis terminal.
7.
Improved postscript print
dialog in View-panel menu.
8.
Histogram client API supports
conversion into Radware format.
9.
Go4 histogram server supports
float histograms.
10.
Execution of ROOT interpreter
commands / macros in the analysis task possible by command line in analysis terminal
window.
11.
Re-design of condition editor:
a.
Display all conditions of array
in different colors or hide them optionally. Visibility in editor is property
of TGo4Condition and stored in auto-save file.
b.
Working view-panel pad and
reference histogram of condition may be changed at any time.
c.
Clear counters button applies
clearing to analysis condition immediately and refreshes editor from analysis.
d.
Statistics inside window
condition limits (integral, maximum, mean, rms, etc) are calculated; these
values are displayed in editor and may be drawn in labels on working pad.
Methods to calculate statistical quantities belong to TGo4WindowCondition class and may be used in analysis, too.
e.
Cursor panel with crosshair
mode and optional marker to pick values from displayed histogram. Cursor may be
set by mouse click, by moving the graphical marker object, or by defining
cursor position in the text fields. Cursor values may be drawn in label on
working pad
f.
Extension of polygon condition
/TCutG is calculated and shown like the borders of the window condition.
g.
Improved creation of new TCutG functionality. Assignment to current polygon condition may be cancelled.
Handles pads with multiple TCutGs.
12.
Added class TXXCalibPar to Go4Example2Step. Shows a procedure how to calibrate spectra using the Go4 fitter in
connection with the parameter mechanism and an ASCII file “database” of line
energies.
13.
Make full screen default for
new view panels.
14.
When updating objects in Memory
folder, a redraw is done automatically.
15.
When monitor updates a
View-panel, the pads are updated without blocking the GUI (not yet for picture)
16.
Button besides zoom buttons to
enter display limits by values
17.
Drag pictures from Analysis pad
to View-panel (only empty view panel, or is inserted in pad)
18.
Some buttons on the browser
pads have been rearranged to be consistent. On Memory browser pad the icons for
"update local objects" and "synchronize with directory"
have been exchanged to be consistent with Analysis pad.
1.
TGraph objects can be registered and displayed correctly. Reset of TGraph (clear all points) by “eraser” button from GUI possible.
2. Reset/clear complete folders by selecting them in remote browser and
“eraser” button. New method ClearObjects(“Histograms”) to reset all objects of
named folder, e.g. all histograms at once.
3. “Print” button to printout histogram and condition lists with
statistics in analysis terminal. These buttons are located in the dynamic list
editor.
4.
Parameter classes may contain TGo4Fitter* references or arrays of these. Fit GUI can be used to edit fitter
from within parameter editor. Framework provides new class TGo4FitterEnvelope as example parameter. Example put into TXXXAnalysis.
5.
User defined event source is
possible. New class TGo4UserSourceParameter to be checked in analysis step factory for any kind of input.
Example package Go4ExampleUserSource shows
usage.
6.
New class TGo4Picture to define layout of canvas with histograms. Pictures are registered
in Go4 Pictures folder and stored in auto-save file like histograms; they can
be displayed in any view-panel. Example added in TXXXAnalysis.
7.
Possibility to register
complete TCanvas objects in Go4 Canvases folder to be saved within auto-save file.
Switch TGo4Analysis into ROOT batch mode to suppress drawing actions in analysis client
while canvas is set up.
8.
Go4 GUI can display and compare
objects from different files in the same view panel now.
1.
Possibility to select rsh or
ssh and analysis output in Xterm or GUI window.
2.
Wildcard in input lmd file
names.
3.
Input file name beginning with
@ is interpreted as text file containing lmd file names.
4.
An auto-save file can be
written on demand (button in configuration menu).
5.
Parameter editor. User
parameter objects (subclasses of TGo4Parameter) registered in the analysis can be edited in the GUI by double
click in the browser. Currently supported members are the primary data types
and arrays of these.
6.
New environment variable
7.
Dynamic lists. A dynamic list
editor can be used to create/specify dynamic entries. A dynamic entry consists
of a histogram (can be created new) and a member of an event object which shall
be histogrammed. Optionally a condition can be added. The condition also can be
created new. The event structure is expanded in the browser. Drag&drop is
provided to select members.
8.
The condition editor has been
improved. Arrays are now handled properly. TCutGs for polygon conditions
can be created new.
9.
TGraph objects are supported like histograms.
10.
In the Go4 view panel, the ROOT
"event status" (cursor position) can be displayed.
11.
The new fit GUI is available.
It includes three different peak finders, a simple fitter, a wizard, and full
access to all fitter components. Fitters can be stored/retrieved to/from files
or memory.
12.
User Makefile: the user
executable need to be linked against the make file variable $(
The Go4 (GSI
Object Oriented On-line-Offline) Analysis Framework has been developed at GSI.
It is based on the ROOT system of CERN. Therefore all functionality of ROOT can
be used.
Go4 has two
parts: the analysis framework itself and a Qt based GUI. Both can be used
independently, or together. The separation of the analysis and GUI in two tasks
is especially useful for on-line monitoring. The analysis runs asynchronously
to the GUI which is (almost) never blocked. The same analysis can be run in
batch/interactive mode or in remote GUI controlled mode. The GUI can be used
stand alone as ROOT file browser and as histogram viewer for GSI standard
histogram servers like MBS. Moreover, the analysis task can be run either as a
client bound to one GUI (default), or can be started as an analysis server with
the possibility to connect several GUIs (one controller and arbitrary number of
observers with restricted commands).
gui150
The Go4
framework handles event structures, event processing, and event IO. The
analysis event loop is organized in steps: Each step has an input event, an output event, and an
event processor. The output event calls the event processor to be
filled. The event processor has also access to the input event. In the current
design the analysis is data driven. A first event object (input1) is filled
from some event source (input). An output event object (output1) is filled by
an event processor object (process1) which has access to both, input1 and
output1. Optionally the output event may be written to a file (file1). In the
next step the input event object (input2) can be either the output event object
(output1) from the previous step or retrieved from the file. The second output
event object (output2) is filled by the second event processor object
(process2) and can be optionally written to a second file.
The information
needed to create the event and processor objects (which are deleted when the
event loop terminates) is stored in step factories which are kept in the
analysis.
The processor
and output event classes have to be provided by the user. The input classes for
standard GSI event sources are provided by Go4 (see chapter 4, page 21). Analysis and step factory classes are provided by
Go4 or can be implemented by the user as subclasses.
gui148
For normal
operation, the Go4 analysis steps are designed to run subsequently. But in
addition, each analysis step has access to the output events of all other
previous analysis steps, so it would be possible to let analysis steps
logically run “in parallel”, all starting with the output event of the first
step, and all delivering their results to the last step that may collect and combine
them.
gui149
Outside the
analysis steps the user functions UserPreLoop(),
UserPostLoop(), and
UserEventFunc() located in the user analysis class
are executed as shown in the figure. In principle, they could be used to
implement the full analysis without using the step mechanism. But for setting
up a new analysis the use of steps is strongly recommended, because steps can
be controlled by the GUI and offer event and IO management.
In the event loop,
after processing the steps and UserEventFunc() the Go4 dynamic list processor is executed. This processor can be dynamically
configured from the GUI to check conditions and/or fill histograms.
gui146
The Go4
concept consists of base classes (interfaces) for event structures, algorithms,
and IO, which can be implemented by user subclasses or by framework plug-ins
(general service classes) delivered with Go4. Class descriptions and reference
guides are available on the Go4 Website http://go4.gsi.de.
The interface classes provided by Go4 (a detailed description is in
the reference manual) are normally not seen by the user. Starting with the
examples (see chapter 5, page 30) one can better study derived working classes.
TGo4EventElement: Defines the event structure and methods to clear and fill this
structure. Input and output event structures of each step of the analysis are instantiated
once at initialization. In the event loop the virtual methods Fill() and Clear() are used to update the event data. These methods can be implemented
in the user subclass. By default, Fill calls BuildEvent of event processor.
TGo4EventSource: The source of the event data. This can be e.g.
a file of a certain format, or a socket connection to an event server. Usually,
the event source class has a BuildEvent(TGo4EventElement*) method, e.g., which
can be called by the Fill() method of the event object to be filled with
the data. Therefore, event element and event source implementation classes have
to “know” each other to perform a matching fill procedure. The class
constructor should open (connect) the source; the destructor should close (disconnect)
it properly.
TGo4EventStore: An object responsible for storing the event
data. This can be e.g. a local file of a certain format, but may as well be a
connection to some storage device. The virtual method Store(TGo4EventElement*) is used to store the pointed event object. The class constructor
should open the storage; the destructor should close it properly.
TGo4EventProcessor: An object that contains the algorithm to convert an input event object
into an output event object (both of class TGo4EventElement). This is a subclass of TGo4EventSource, since it delivers the filling of the output event from the input
event. The event processor implementation has to “know” the input and output
event classes. The methods of converting the data (i.e. actually performing the
analysis) are free to be defined by the user. By default a BuildEvent method is provided.
TGo4EventFactory: Defines the actual implementations of all the
above. Go4 uses a factory design pattern to create all event class objects at initialization.
The virtual methods:
CreateInputEvent(), CreateOutputEvent(), CreateEventSource(TGo4EventSourceParameter*),
CreateEventStore(TGo4EventStoreParameter*),
CreateEventProcessor(TGo4EventProcessorParameter*) have to be defined in the user factory. They create the
respective objects and return the pointer to it. The default factory provides
methods
DefEventProcessor(objectname, classname),
DefInputEvent(objectname, classname) and
DefOutputEvent(objectname,classname).
Simple examples of a running Go4 analysis can be found on
directories $
Go4 offers predefined implementations of the event base classes,
including an interface to the GSI data acquisition Multi Branch System MBS, the
GSI list-mode files, and ROOT files.
TGo4EventElement (base
class):
TGo4MbsEvent,
TGo4MbsSubEvent
MBS event format 10-1
TGo4CompositeEvent Base class for all composite event structures
TGo4ClonesElement Clonesarray container for composite event
TGo4EventSource (base class):
TGo4MbsFile
(read
from *.lmd list-mode file with format 10,1)
TGo4MbsEventServer (connect to MBS event server)
TGo4MbsStream (connect
to MBS stream server)
TGo4MbsTransport (connect to MBS transport server)
TGo4RevServ
(connect to remote event server)
TGo4FileSource (read from *.root file from Go4 tree, i.e. one
file containing one TTree per analysis step)
TGo4EventStore (base
class):
TGo4FileStore (write to *.root file with Go4 tree, this file
can be used as TGo4FileSource later)
TGo4BackStore Use
TTree existing only in memory to view and analyze event structures.
These classes can be used directly to write
simple analysis.
Using these implementations,
getting MBS event data into ROOT (without Go4 framework) could look like this:
#include
"Go4EventServer/Go4EventServer.h"
#include
"Go4Event/TGo4EventEndException.h"
int
main() {
TGo4EventSource* input = new TGo4MbsFile("file.lmd"); // MBS
list-mode file
// TGo4EventSource* input= new TGo4MbsTransport("node"); // MBS transport server
// TGo4EventSource* input= new TGo4MbsStream("node"); // MBS stream server
// TGo4EventSource* input= new TGo4MbsEventServer("node");
// MBS event server
// TGo4EventSource* input= new
TGo4RevServ("node"); //
Remote event server
TGo4EventStore* output = new
TGo4FileStore("output",1,5);
// split level, compression
TGo4MbsEvent* event = new TGo4MbsEvent();
event->SetEventSource(input);
event->Init();
Int_t eof = 0, numEvents = 0;
while(eof==0) {
try{
event->Fill(); // read
event
numEvents++; // eof
throws exception
output->Store(event); // write to
file
}
catch(TGo4EventEndException& ex) {
eof=1; } // mark end of file
catch(...) { cout <<
"Error" << endl; eof=2; } // any other error
}
cout << "EOF after "
<< numEvents << " events" << endl;
}
The events in the ROOT
file can be retrieved by program, but not in tree viewers. For the use of tree
viewers, a new output event object should be filled and stored.
As
mentioned above a Go4 analysis is organized in steps. The information needed to
instantiate a step is kept in the step factory.
TGo4EventFactory (base
class):
TGo4EventServerFactory (base class): (contains
factory methods that already know the above implementations. User step
factories must inherit from this class!)
TGo4StepFactory This
TGo4EventServerFactory can be used in most cases as user factory to
set up the analysis steps (examples Simple and 1Step).
TGo4AnalysisStep objects of this class hold the definition of an analysis
step.
Each analysis
step has at least an input event object, an output event object and an event
processor object. Additionally, it can have an event source (e.g. TGo4FileSource) and an event store (TGo4FileStore) instance. An analysis step is set up by a TGo4EventServerFactory subclass, i.e. TGo4StepFactory or a user defined subclass.
Once the user has defined his/her event class implementations, the
analysis steps can be created and registered to the Go4 analysis framework. The
actual framework consists of the TGo4Analysis class, which is a singleton (i.e. there is only one framework
object in each process). This class provides all methods the user needs, it
keeps and organizes the objects (histograms,...), it initializes and saves the
data objects.
The user
analysis is set up in a subclass of TGo4Analysis, i.e. TUserAnalysis. Constructor and destructor of this user class, in addition with
the overridden virtual methods UserEventFunc(),
UserPreLoop(), and
UserPostLoop() specify the user analysis.
If the latter
functions are not needed, one can also use the TGo4Analysis class directly, as shown in the
example Simple.
In the
constructor of the TUserAnalysis class the analysis step objects are created, each containing
instances of its user step factory. The analysis steps are registered at the TGo4Analysis framework, input and output events of subsequent steps are checked
for matching. Furthermore, other objects like histograms, conditions or
parameters can be created in the constructor and registered, so the framework
is responsible for their persistence. Such objects can also be created in the
step processor.
In addition to
the event processors, the UserEventFunc() allows the user to specify analysis operations that are called once
in each analysis cycle, e.g. filling certain histograms from the output events
of all analysis steps. The UserEventFunc() makes it even possible to call an external analysis framework event
by event without using the Go4 Analysis Steps at all, thus taking advantage of
the Go4 object management and remote GUI features.
The UserPreLoop() and UserPostLoop() functions may define actions that are executed before starting, or
after stopping the main analysis loop, respectively.
Once the user
analysis class is defined, there are two modes of operation: The
single-threaded batch mode, and the multi-threaded client mode that connects to
the Go4 GUI.
The constructor of a TGo4Analysis derived
user class could create one analysis step with input from an MBS file with the
following code fragments (note that we use the standard Go4 step factory
class):
TUserAnalysis::TUserAnalysis()
{
//...
TGo4MbsFileParameter* input;
TGo4StepFactory* factory; // standard factory provided by Go4
TGo4AnalysisStep* step;
input =
new TGo4MbsFileParameter("file.lmd");
factory =
new TGo4StepFactory("Factory");
step =
new TGo4AnalysisStep("Analysis",factory,input,0,0);
// the objects specified here will be
created by the framework later:
factory->DefEventProcessor("XXXProc","TXXXProc");//
object name, class name
factory->DefOutputEvent("XXXEvent","TXXXEvent");
// object name, class name
step->SetSourceEnabled(kTRUE);
step->SetProcessEnabled(kTRUE);
AddAnalysisStep(step);
//...
}
// Example of using the event loop
functions for a trivial counting of events
// fEvents must be defined in TUserAnalysis.h:
Int_t
TUserAnalysis::UserPreLoop() {
fEvents=0;
return
0;
}
Int_t TUserAnalysis::UserEventFunc() {
fEvents++;
return
0;
}
Int_t
TUserAnalysis::UserPostLoop() {
cout << " Total events: " <<
fEvents << endl;
return
0;
}
Typically the user provides the main
analysis program. One can use one of the examples. The main program sets up the
analysis. Then it starts two different modes (see example below):
In batch mode,
the constructor of the user analysis class (e.g. TUserAnalysis or TGo4Analysis) creates the
framework.
The InitEventClasses() method uses the factories of all steps to create the event classes
and open the event sources, event stores, etc. It also checks for consistency
of subsequent steps.
The RunImplicitLoop(Int_t
n) calls the implicit event loop and runs the main
analysis cycle (event processing of all enabled steps, UserEventFunc()) for n times.
CloseAnalysis()
deletes all event classes and closes all
input/output files and connection. This method is complementary to InitEventClasses() that creates them.
The destructor
of the user analysis class calls CloseAnalysis(); in addition the auto-save file is closed and the complete framework
is shut down.
In the
interactive GUI mode, the analysis framework is created with the user analysis
class object, as for the batch mode. Additionally, the framework is handed over
to a TGo4AnalysisClient object that manages the connection to the GUI. Usually, the Go4 GUI
is started first and launches the analysis framework in a remote shell. The
user analysis program is called in the shell script
AnalysisStart.sh in the user's working directory. The working directory as well as
the name of the executable is passed from the GUI side. Then the user
executable creates the analysis framework and connects the multi-threaded
analysis client to the Go4 GUI. After the connection is established, the
complete analysis framework can be controlled from the GUI. After the example,
we describe in detail what is happening on startup of the analysis client and
what effect the GUI control actions have.
As a default, the TGo4AnalyisClient object will set up the analysis-GUI connection in a way that the analysis is a single client to a single GUI as server, as described in section 4.5.2. However, it is possible to run the analysis as a server to connect many GUIs (one controlling GUI and many observer GUIs). Still the analysis class object is handed over to the TGo4AnalysisClient object, but the analysis “client” may run in a network server mode by constructor parameter (note: the classname TGo4AnalyisisClient was not changed for backward compatibility , although it should rather be called TGo4AnalysisSlave to point out the role as command receiving entitiy).
The analysis server may be started independently from the GUI from a shell like in the batch mode, and may already start analysis run from preferences setup without any controlling GUI. However, it can as well be launched from the StartAnalysis dialogue of the Go4 GUI when selecting the mode “as server”. A Go4 GUI is ready to connect any such started server. Login of GUI to the analysis server may be with observer, controller, or administrator role, respectively; their passwords can be set in user analysis code. There can be only one controller or administrator, but multiple observer GUIs. Observers may only view exisiting objects, but may not modify them or change analysis setup and running state. Controller may view and modify objects and analysis configuration, but is not allowed to terminate analysis server. Only Administrator may shutdown the analysis server, too.
Additionally, the analysis server may be launched first from one GUI in an xterm, and then connected from this GUI and other GUIs later on. See section 6.3.2 for more details on connection of the GUI client.
The following examples show the essential
structures to start/run the analysis. See also the running examples.
Using user event and processor classes,
but others all Go4 standard:
TROOT go4application("
int main(int argc, char **argv) {
TApplication theApp("Go4App", 0, 0); // ROOT application loop
// ...
TGo4Analysis* analysis = new
TGo4Analysis::Instance();
TGo4StepFactory* factory = new TGo4StepFactory("Factory");
TGo4AnalysisStep* step = new
TGo4AnalysisStep("Analysis",factory,0,0,0);
analysis->AddAnalysisStep(step);
step->SetEventSource(new TGo4MbsFileParameter("myfile.lmd"));
// tell the factory the names of
processor and output event
// both will be created by the
framework later
// Input event is by default an MBS
event
factory->DefEventProcessor("XXXProc","TXXXProc");//
object name, class name
factory->DefOutputEvent("XXXEvent","TXXXEvent");
// object name, class name
if (strcmp(argv[1],"-gui") == 0) // was started from GUI:
create analysis client
TGo4AnalysisClient* client = new TGo4AnalysisClient(argc,argv,analysis,kFALSE,"","");
else { // run implicit event loop
analysis->InitEventClasses();
analysis->RunImplicitLoop(100000); // number of events
delete analysis;
gApplication->Terminate(); // exit
}
theApp.Run(); // needed to run the client, or if gApplication was not
terminated
} // end main
Using user analysis, event and processor
classes (steps set up in TUserAnalysis):
TROOT go4application("
int main(int argc, char **argv) {
TApplication theApp("Go4App", 0, 0); // ROOT application loop
// prepare arguments for TUserAnalysis
// ...
TUserAnalysis* analysis = new TUserAnalysis(...); // arguments as required
if (strcmp(argv[1],"-gui") == 0) // was started from GUI:
create analysis client
TGo4AnalysisClient* client = new
TGo4AnalysisClient(argc,argv,analysis,kFALSE,"","");
else { // run implicit event loop
analysis->InitEventClasses();
analysis->RunImplicitLoop(100000); // number of events
delete analysis;
gApplication->Terminate(); // exit
}
theApp.Run(); // needed to run the client, or if gApplication was not
terminated
} // end main
Objects used in Go4 are organized in ROOT folders. The folder
structure is sent to the GUI. Objects must be registered in the analysis to be
seen in the GUI browser. Registered objects can be located in the processors.
The top folders as seen in the GUI are shown on the left side. The methods to
register/locate objects are (pointer to the appropriate object, optional
subfolder as string, name including subfolder as string):
·
AddHistogram(pointer,subfolder),
GetHistogram(name)
·
AddAnalysisCondition(pointer,subfolder),
GetAnalysisCondition(name)
·
AddParameter(pointer,subfolder),
GetParameter(name)
·
AddPicture(pointer,subfolder),
GetPicture(name)
These methods
are available in TGo4Analysis and TGo4EventProcessor subclasses. Objects created in a TGo4Analysis subclass can be located in all event processors. Objects created in
event processors can be located in all subsequent event processors (steps).
Registered
objects are stored/retrieved to/from the auto-save file, if enabled. Retrieval
is done after creation of the analysis singleton before the
creation of the steps. When an object retrieved from the auto-save file is created
in a processor the retrieved object is replaced (stored data lost). When an
object is created in the analysis singleton it will be replaced by the one
retrieved from the auto-save file except histograms which are not retrieved in
this case. This means that histograms created in the analysis singleton are
always empty after startup.
Parameters used
in the analysis are implemented by the user in classes derived from TGo4Parameter. Such objects
are registered to the framework and can be edited by a generic parameter editor (see chapter 6.11.2, page 74). Parameter objects can be created in the user
analysis or the event processor class. Parameter objects are loaded from an
optional auto-save file after instantiation of
the analysis and before instantiation of the processor objects. When created in
the analysis the values set in the constructor are therefore overwritten by
auto-save. To use the GUI editor, the UpdateFrom() method must be implemented to update the local (active) parameter
object from the modified one delivered by the editor. In this method it is up
to the user to ignore certain members or to execute whatever he wants. E.g. one
could use parameters to execute commands. Parameters in the auto-save file can
be edited. In the editor they can be saved/retrieved to/from files. Several
mechanisms can be implemented to handle the parameter member values. The main
question is how restricted the methods of modification should be.
Conditions are objects holding window limits or polygons. One or two
values can be checked against the limits or the polygon, respectively. In
addition the conditions have test and true counters. They can be set to return
always true or false or return the inverted test result. They can be edited by
the GUI (see chapter 6.8.2, page 67). They can be used to steer the analysis flow. They
are saved/retrieved to/from the auto-save file, if enabled. They can be edited
in the auto-save file. In the editor they can be saved/retrieved to/from files.
If a mechanism like for the parameters (4) is wanted, one can use macro savecond.C([file],wildcard,prefix) from $
Example:
root[0].x savecond.C("myfile.root","*","setcon")
would produce
macros setcon_cond1.C, setcon_cond2.C etc. The macros have three arguments: restore flags, restore
counters, reset counters (0=no, 1=yes), e.g. setcon_cond1(1,0,1).
When
starting the Go4 analysis in GUI mode, the following actions take place in that
order:
1.
The Launch
Analysis GUI panel started by reads some settings from file $
AnalysisStart.sh
in the current directory of the user analysis. Here the user may add his own initializations.
2.
In “as client” mode (default), the executable MainUserAnalysis is started in GUI mode with the parameters:
MainUserAnalysis –gui <analysis name> <host name> <port number>
The parameters analysis name and host name are taken
from the launch client GUI panel, the port number is dynamical (displayed on
GUI start-up in the shell, often=5000). Instead of launching the client from
the GUI, one may start the analysis client manually from a separate shell (do
not forget “.
go4login” !) with the matching host name and port number for the running Go4
GUI. This can be useful if the analysis shall run under gdb, or if
ssh/rsh fails for some reasons.
Started in “as server” mode, the analysis executable (e.g. in Go4Example2Step)
is called with arguments parameters
MainUserAnalysis –server <analysis name>
and starts in server mode (see also 6.3.2, page 46).
3.
TGo4Analysis is created and initializes the analysis framework. Then the
constructor of the user subclass (e.g. TXXXAnalysis) defines the list of analysis
steps with initial event parameters (input and output filenames) and auto-save
settings just as passed from the MainUserAnalysis. Additionally, some
user objects may be created and registered here. Note that histograms
registered here are never updated or replaced from the Go4 auto-save file and
exist only until the analysis client is terminated. Conditions and parameters,
however, are updated when the auto-save file is loaded and if their name is existing
there.
4.
The analysis slave, if in client
mode, connects to the Go4 GUI. Optionally, the Go4 histogram/object
server is created.
Note that the analysis in server mode does not connect automatically to the
starting GUI, but waits for a separate connect request with login and password
from any GUI. Only after this explicit connection the GUI in -client
mode gets control over the analysis server!
5.
The analysis settings are
loaded from the default preferences file Go4AnalysisPrefs.root. A message is sent to the GUI: “Analysis Client MyClient: Status Loaded from file Go4AnalysisPrefs.root ” (if successful). Note
that all settings specified before in the compiled code (auto-save file name,
event sources, etc.) are overwritten if the preferences file exists.
6.
The analysis objects are loaded
and updated from the auto-save file. The file name from the loaded analysis
settings is used, if existing. Otherwise, the filename specified in the preceding
user code by SetAuto-saveFile(Text_t* name) is used. If successful, a message is sent to the GUI: “Analysis Client MyClient: Objects Loaded.” If auto-saving was disabled
completely by calling
SetAutoSave(kFALSE), the
auto-save file is not opened here even if it exists, and no objects are loaded!
The “overwrite filename” option in the auto-save settings must be disabled
to recover objects of a previous auto-save file; otherwise, all objects in an
old file of the same name are lost!
7.
The analysis settings are
displayed on the GUI. At this moment, the analysis configuration window pops up
and shows the active settings. Note that for GUI client at analysis server,
configuration does not pop up automatically after login, but has to be requested
by “arrow right” button of analysis configuration window.
8.
End of analysis start-up. A
message is sent to the GUI: “Analysis Client MyClient has finished initialization”.
Note that
now the analysis itself is not yet initialized, i.e. the event objects have not
been created, and there are still no connections to event sources, etc.
At any
time the user may apply new settings to the analysis and start/stop the run. Note
that if GUI runs as client connected to an analysis server, these operations
are permitted for controller or administrator login only. Here the following is
happening in the described order:
1.
Submit the analysis settings. The
settings as displayed in the analysis configuration window are sent to the analysis
client.
i.
First, an already existing analysis is closed
(see below).
ii.
The analysis is initialized
with the new settings. Objects are loaded from the new auto-save file except
auto-save is disabled by SetAutoSave(kFALSE). The file name is as specified in the configuration window.
iii.
The event objects are created.
Event sources and stores are opened. The constructors of all user events and event
processors are executed. Note that any object (histogram, parameter, etc)
which is created and registered in the user event constructors might replace an
object of same name that was loaded from the auto-save file before! To
continue working with the loaded objects, the user should request pointer to
the object by name from the framework here. Only if the object was not found it
should be created anew.
After
submit, the Analysis browser can be refreshed by . When an analysis was running before, the new
analysis is started immediately and the refresh is done automatically.
2.
Start the analysis with :
i.
The Go4 GUI will send the start
command and refresh the view in the analysis browser.
ii.
The UserPreLoop() function is executed once. Here transient pointers to data might be
initialized, values from a user file might be read, etc.
iii.
The Analysis event loop is
starting. For each event the analysis steps, the dynamic list entries, and the UserEventFunc() are executed. The loop will run until the event source is at the
end, an error occurs, or the stop command is applied by the user.
3.
Stop the analysis with :
i.
The event loop is halted. This
will not close the analysis itself, i.e. all event objects still exist, event
sources and -stores are still open. When restarting the analysis by ,
it will continue with the next event.
ii.
The UserPostLoop() function is executed once. Here transient pointers should be reset
to 0, user files might be written or closed, etc.
4.
Save configuration settings: At any time the current settings can be
saved to a preferences file. This will not affect the running analysis. Note
that after changing the settings in the analysis configuration window they must
first be submitted to save them!
5.
Load Settings: Loading analysis settings from a preferences file will
immediately close the running analysis. The closing actions are just as
described below. However, the loaded settings are not initialized until they
have been Submitted again from the analysis
configuration window!
The
analysis client is shut down with the a button. This will take the following actions
1.
The connection between analysis
and GUI is closed.
2. The destructor of the user analysis class is executed.
3.
Close of the analysis:
i.
Objects are written to the previous auto-save file, if SetAutoSave(kTRUE).
ii.
The event objects are deleted.
Go4 event sources (.lmd files and MBS connections) are closed. Event stores
(.root files) are finally written and closed. The destructors of all user events
and event processor classes are executed. All references to the event objects
are deleted from the Go4 folders.
iii.
The dynamic list is reset. All
pointers to non existing objects are cleaned up.
4. The analysis client executable terminates. The Go4 GUI is ready to connect the next analysis client.
The GUI may disconnect the analysis server with the button. This will neither stop the analysis nor shutdown the server task, but just close the connections to this GUI. Additionally, when connected to an analysis server, the GUI has a button in the analysis toolbar and a menu for Shutdown Analysis server. This is permitted in administrator mode only! This will take the following actions:
1.
Analysis server broadcasts message
about shutdown to all GUI clients connected. The GUIs will cease monitoring
activities and prepare for disconnect.
2.
The destructor of the user
analysis class is executed.
3.
Close of the analysis, see details in 4.5.10
4.
The analysis server disconnects
all GUI clients fast, i.e. without handshaking protocol, and terminates.
To begin with
Go4, there are examples of analysis packages at $
Example |
Analysis |
Step
factories |
Event
objects |
Steps |
Simple |
TGo4Analysis |
TGo4StepFactory |
TGo4EventElement |
Analysis |
1Step |
TXXXAnalysis |
TGo4StepFactory |
TXXXEvent |
Analysis |
2Step |
TXXXAnalysis |
TXXXUnpackFact |
TXXXUnpackEvent |
Unpack |
Mesh |
TMeshAnalysis |
TGo4StepFactory |
|
13 different |
When using Go4 at GSI where it is already
installed, Go4 is set up by
Note that there must be a space behind the
dot. To see all relevant environment variables use command
The output of this command would be helpful
if you report problems.
Copy the content of the directory $
The package consists of the following files
besides the include and source files:
·
Readme.txt
·
AnalysisStart.sh is the standard startup
script for analysis client One may add here definitions needed by the analysis.
·
Makefile
·
Go4UserAnalysisLinkDef.h contains ROOTCINT class pragma definitions
·
rename.sh is a script to set up file/class names.
Before renaming the files, cleanup by
command:
There is one string included in all class
and file names: XXX. It is recommended to replace this by another string more
specific. This is done by
Example:
Note that "myname" will be
part of all class and file names! Hint: do
not use a string which is already in any filename!
Then rebuild the package by command
Shared library libGo4UserAnalysis.so and executable MainUserAnalysis should be created.
rsh
When the analysis program is started from
the GUI, one can choose between rsh and ssh. For rsh shell, make sure that the
file .rhosts exists and that it contains
entries for the machine names you want to run the Go4 analysis client on. The
file .rhosts
could e.g. look like this:
node01
node02
localhost
Note: localhost should be listed here,
since this is the Go4 default.
ssh
To use ssh one must create ssh keys. These
keys are specific to the node where they are created:
cd
~/.ssh
ssh-keygen
-d
This package on Go4ExampleSimple contains a simple running Go4 analysis. It contains one analysis
step. It uses the standard Go4 analysis classes TGo4Analysis, TGo4StepFactory and
TGo4EventElement. Therefore the functions UserPreLoop(),
UserPostLoop(), and
UserEventFunc() are not available. No data can be
stored in the output event. The example uses some conditions and some parameter
objects. The step is reading events from a standard MBS event source filling
some histograms. No output file is written. The analysis processes up to eight
long word values from up to two sub events. A suited input file can be found on
the Go4 web or “MBS Random” event source can be used. All classes are defined
and declared in two files (*.h and *.cxx). Additional descriptions are in the
source files.
The main program
can be started from the Go4 GUI (see chapter 6.3, page 45) or by command line:
./MainUserAnalysis
-file|-trans|-stream|-evserv|-revserv input [-p port] [events]
./MainUserAnalysis -f myfile.lmd
The events can
be read from standard GSI lmd files or MBS or event servers. For each event the
user event processor TXXXProc (function BuildEvent) is called. This user event processor fills some histograms.
The macro MainUserAnalysisMacro.C can run directly in ROOT. It needs a .rootmap
file for automatically loading all necessary libraries. This file is created by
the new files Makefile and Module.mk
from the example.
The analysis,
analysis factory, and analysis step (all standard Go4 classes) are created in
the main program. The input is specified by a set of macros ( file.C, trans.C, stream.C, evserv.C,
revserv.C). Other setups are done in macro setup.C. The macros
are called in the main program by gROOT->ProcessLine.
The event
filled: TGo4EventElement (no
data, no user specific code)
The processor: TXXXProc
The standard
factory created in the main program keeps all information about the step. No
user event class is used in this example. Members of TXXXProc are histogram, condition, and parameter pointers used in the event
function BuildEvent(). In the constructor of TXXXProc the histograms, parameters and conditions are created. Method
BuildEvent() - called event by
event - gets the output event pointer as argument, but do not fill any data.
The input event pointer is retrieved from the framework. In the first part,
data from the raw input MBS event are copied to arrays of TXXXProc. Two sub-events (crate 1,2) are processed. Then the histograms are
filled, the 2d one with polygon conditions.
Parameter class TXXXParam
In this class
one can store parameters, and use them in all steps. Parameters can be modified
from GUI.
See also chapter
6.4.4, page 49. By default auto-save is enabled for batch, disabled
with GUI. The name of the file is built from the input by
<input>_AS.root
If it is enabled all objects are saved into this ROOT file at the end of the event loop. At startup the auto-save file is read and all objects are restored from that file. When TGo4Analysis is created, the auto-save file is not yet loaded. Therefore the objects created here are overwritten by the objects from auto-save file (if any), except histograms. From GUI, objects are loaded from auto-save file when the Submit button is pressed. Note that histograms are not cleared. One can inspect the content of the auto-save file with the Go4 GUI. Note that appropriate user libraries should be loaded into GUI to access data from auto-save file (see chapter 6.2, page 45).
All lines with
**** are from the example classes.
>
MainUserAnalysis –event MBS42 1000
**** Input MBS42 (-e)
process 1000 events
auto save file: MBS42_AS.root
**** evserv.C: Create MBS event server
input MBS42
**** setup.C: Setup analysis
****
**** Factory: Create input event for MBS
**** Factory: Create event processor
XXXProc
**** TXXXProc: Create instance XXXProc
**** TXXXProc: Restored histograms from
autosave
**** TXXXProc: Restored conditions from
autosave
**** TXXXProc: Restored pictures from
autosave
**** Factory: Create output event
XXXEvent
**** Event XXXEvent has source XXXProc
class: TXXXProc
**** TXXXProc: Delete instance
****
Creating a new
class
Provide the
definition and implementation files (.h and .cxx)
Add class in Go4UserAnalysisLinkDef.h
Then make all.
Most probably you
will change TXXXParam to keep useful parameters.
Then definitely you will change TXXXProc to create your histograms, conditions, pictures, and finally write your analysis function BuildEvent().
Before running MainUserAnalysisMacro.C in ROOT CINT it must be changed because all parameters are hard coded.
This package on Go4Example1Step contains a simple running Go4 analysis. It contains one analysis
step. It uses the standard Go4 step factory TGo4StepFactory, but a user written
TXXXAnalysis. In this class the functions UserPreLoop(),
UserPostLoop(), and
UserEventFunc() can be used. It uses some
conditions and some parameter objects. The step is reading events from a
standard MBS event source filling some histograms and an output event. The
analysis processes up to eight long word values from up to two sub events. A
suited input file can be found on the Go4 web. All classes are defined and declared
in two files (*.h and *.cxx). Additional descriptions are in the source files.
Setup: TXXXAnalysis
The main program
can be started from the Go4 GUI (see chapter 6.3, page 45) or by command line:
./MainUserAnalysis
-file|-trans|-stream|-evserv|-revserv input [-output] [events]
./MainUserAnalysis -f myfile.lmd
The events can
be read from standard GSI lmd files or MBS or event servers. For each event the
user event processor TXXXProc (method BuildEvent()) is called. This user event processor fills some histograms and an
output event TXXXEvent (raw event) from the input event. The output events can optionally
be stored in ROOT files. When a ROOT file with raw events exists, it can be
viewed by the Go4 GUI using the tree viewer. Note that appropriate library
should be loaded into GUI to let the viewer know TXXXEvent (see chapter 6.2, page 45).
In TXXXAnalysis the analysis step is created with the step factory and input and
output parameters. Here the defaults are set concerning the event IO. Two parameter
objects are created (TXXXParam and TXXXControl).
The event
filled: TXXXEvent
The processor: TXXXProc
The standard
factory created in TXXXAnalysis keeps all information about the step. The TXXXEvent contains the data members to be filled in TXXXProc from the input event (MBS 10,1). The Clear() method must clear all these members (an array for each crate in the
example). The analysis code is in the event processor TXXXProc. Members are histograms, conditions, and parameter pointers used in
the event function BuildEvent(). In the constructor of TXXXProc the histograms and conditions are created, and the pointers to the
parameter objects (created in TXXXAnalysis) are retrieved. Function BuildEvent ()- called event by event - gets the output event pointer as
argument (TXXXEvent). The input event pointer is retrieved from the framework. In the
first part, data from the raw input MBS event are copied to the members of output
event TXXXEvent. Two sub-events (crate 1,2) are processed. Then the histograms are
filled, the 2d one with polygon conditions.
The name of the
optional output file is built from the input by
<input>_XXXEvent.root
Parameter class TXXXParam
In this class
one can store parameters, and use them in all steps. Parameters can be modified
from GUI.
Parameter class TXXXControl
This class has
one member "fill" which is checked in TXXXProc->Event() to fill histograms or not. The macro
setfill.C(n), n=0,1 can be used in the GUI to switch the filling on or off. It
creates macro histofill.C() which is actually used to set filling on or off (in TXXXProc). You can also modify histofill.C by editor before running
the analysis.
See also chapter
6.4.4, page 49. By default auto-save is enabled for batch, disabled
with GUI. The name of the file is built from the input by
<input>_AS.root
If it is enabled
all objects are saved into this ROOT file at the end of the event loop. At
startup the auto-save file is read and all objects are restored from that file.
When TXXXAnalysis is created, the auto-save file is not yet loaded. Therefore the
objects created here are overwritten by the objects from auto-save file (if
any), except histograms. From GUI, objects are loaded from auto-save file when
the Submit button is pressed. Note that histograms are not cleared. One can
inspect the content of the auto-save file with the Go4 GUI. Note that appropriate
user libraries should be loaded into GUI to access data from auto-save file
(see chapter 6.2, page 45).
All lines with
**** are from the example classes.
> MainUserAnalysis –e MBS42 1000
**** Input MBS42
(-e)
output MBS42_XXXEvent.root disabled
process 1000 events
auto save file: MBS42_AS.root
**** Analysis: Create MBS event server
input MBS42
****
**** TXXXParam Par1 updated from auto
save file
**** TXXXControl Control updated from
auto save file
**** TXXXControl: Histogram filling
enabled
**** Factory: Create input event for MBS
**** Factory: Create event processor
XXXProc
**** TXXXProc: Create instance XXXProc
**** TXXXControl: Histogram filling
enabled
**** TXXXProc: Restored histograms from
autosave
**** TXXXProc: Restored conditions from
autosave
**** TXXXProc: Restored pictures from
autosave
**** Factory: Create output event
XXXEvent
**** TXXXEvent: Create instance
XXXEvent
**** TXXXAnalysis: PreLoop
First event #: -1926055269
**** TXXXAnalysis: PostLoop
Last event #: -1926053525 Total events: 1000
**** TXXXAnalysis: Delete instance
**** TXXXEvent: Delete instance
**** TXXXProc: Delete instance
****
Creating a new
class
Provide the
definition and implementation files (.h and .cxx)
Add class in
Go4UserAnalysisLinkDef.h
Then make all.
Most probably
you will change TXXXParam to keep useful parameters.
Then you might
change TXXXEvent to represent your event data.
Keep the Clear() method consistent with the data members!
Then definitely
you will change TXXXProc to create your histograms, conditions,
pictures, and
finally write your analysis function BuildEvent().
In TXXXAnalysis there are three more functions which eventually can be useful:
UserPreLoop () - called before event
loop starts,
UserEventFunc() - called after each TXXXProc::BuildEvent(),
UserPostLoop () - called after event loop stopped.
This example on Go4Example2Step contains an unpack step and an analysis step. It uses some
conditions and some parameter objects. Step one is reading events from a
standard MBS event source filling some histograms and an output event. Step two
uses this event as input and fills another output event and some more
histograms. The analysis processes up to eight long word values from up to two
sub events. A suited input file can be found on the Go4 web.
The main program
(MainUserAnalysis) can be started from the Go4 GUI (see chapter 6.3, page 45) or by command line:
./MainUserAnalysis
-file|-trans|-stream|-evserv|-revserv|-random
input [-server] [-port #][-output]
[events]
./MainUserAnalysis -f myfile.lmd
The events are
read from standard GSI event sources (in the GUI one can switch to MBS or event
servers). Then the first user event processor is called (Unpack). This
user event processor fills some histograms and the first user event (unpacked
event) from MBS input event. Then the second user event processor is called (Analysis).
This user event processor fills some other histograms and the second user event
(calibrated event) from the first event. The events from the first and second
step can optionally be stored in ROOT files (from GUI). When a ROOT file with
unpacked events exists, the first step can be disabled, and this file can be selected
as input for the second step (from GUI).
The main program
builds the files names needed and creates the TXXXAnalysis. Then it either connects to the GUI (when started from GUI) or
starts the event loop (when started from shell). If –server is specified,
GUIs invoked by go4
–client may connect.
Files created by
the example are
Go4AnalysisPrefs.root:
saved preferences
<input>_AS.root:
auto-save file
<input>_XXXUnpack.root:
event tree output from step 1
<input>_XXXAnl.root:
event tree output from step 2
All classes
are defined and declared in two files (*.h and *.cxx)
In TXXXAnalysis the two steps are created with
their factories and input and output parameters. Here the defaults are set
concerning the event IO. Two parameter objects are created (TXXXParameter). They can be used in both steps.
The
factory: TXXXUnpackFact
The event
filled: TXXXUnpackEvent
The
processor: TXXXUnpackProc
The factory TXXXUnpackFact normally need not to be
changed as long as standard GSI event sources are used.
The TXXXUnpackEvent contains the data members to
be filled from the input event (MBS 10,1). Only the Clear() method must be changed to clear all these members.
The unpacking code is in the event processor TXXXUnpackProc. Members are histograms, conditions, and parameter pointers used in the event method XXXUnpack(). This name can be chosen by the user. In the Fill() method of TXXXUnpackEvent this method must be called. In the constructor of TXXXUnpackProc the histograms and conditions are created, and the pointers to the parameter objects (created in TXXXAnalysis) are set. XXXUnpack() - called event by event - gets the output event TXXXUnpackEvent as argument (poutevt). The input event is retrieved from the framework. The first eight channels of crate one and two are filled in histograms Cr1Ch01-08, Cr2Ch01-08, respectively. His1g is filled under condition cHis1 on channel 0, His2g under condition cHis2 on channel 1. When editing conditions cHis1,2 histograms His1,2 filled by channel 0,1 will be displayed automatically to set/display the condition values. Picture condSet shows histograms His1,2 on top, His1,2g at bottom. Open the condition editor in the view panel of the picture. Conditions cHis1,2 will be selectable. They are displayed in the pad where they should be set. Both conditions are attached to the picture (see chapter 6.8.4, page 69). Histogram Cr1Ch1x2 is filled for three polygon conditions: polycon, polyconar[0], polyconar[1], all on the same values as the histogram.
The
factory: TXXXAnlFact
The event
filled: TXXXAnlEvent
The
processor: TXXXAnlProc
The step two is
build in the same way as step one.
Note that the TXXXUnpackEvent is used two times: once as
output of step one, and once as input of step two. Therefore the Fill() method checks if TXXXUnpackEvent has to
be filled by XXXUnpack() in step one or retrieved from input file of step two which should
be an output file of step one. Step one must be disabled in the second case.
The user method XXXEventAnalysis() always gets the pointer to the correct event. Histogram Sum1 is filled by first 4 channels of crate 1 and first 4 channels of
crate 2. All channels are gated with condition wincon1. Histograms Sum2,3 are
filled similar, but without gate, and shifted by XXXPar1,2->frP1. Histogram Sum1calib is
filled like Sum1 without
gate but with values calibrated by method TXXXCalibPar->Energy()
of parameter calipar.
With the TXXXParameter class one can store parameters, and use them in all steps. Parameters can be modified from GUI by double click.
In TXXXCalibPar is an example how to use fitters in parameters to calibrate
histograms (more chapter 6.11.3, page 75).
There are a few conditions created in TXXXUnpackProc. One (polycon) is used in XXXUnpack() for the accumulation of histogram Cr1Ch1x2. Another one (wincon1) is used in XXXEventAnalysis() of TXXXAnlProc to fill histogram Sum1. Conditions can be modified by double click in the browser. One can attach a histogram to a condition or attach conditions to picture pads to ensure that the condition is displayed/set on the proper display.
This example on Go4ExampleMesh shows
how to set up a Go4 analysis of several steps that build a mesh of parallel
analysis branches with different result generations. Additionally, one can see
how the improved TGo4FileSource class supports partial input from a ROOT tree.
The setup of the mesh analysis is done in
the constructor of the TMeshAnalysis class. As in the Go4ExampleSimple, the general TGo4StepFactory is used to specify the event objects by name and class name. An
overall of 13 analysis steps is defined for this example. Generally, the
analysis mesh consists in two different kinds of steps, the execution steps and
the provider steps. The unpack step, however, is as in the other examples just
delivering sample data from a TGo4MbsSource (standard Go4 gauss example).
The step structure of the example mesh is
as sketched in this figure (arrows show dataflow):
These analysis steps do the actual
analysis work, i.e. they convert some input event into the output event. This
is the same as in the more simple examples (2-Step). However, to realize a mesh
structure, the execution steps do not work directly on their own input event as
assigned from the Go4 framework, but use the input event of one or more
provider steps. The execution steps can access the input event pointers of any
provider step by the provider step name, using the GetInputEvent("stepname") method. Note that the native input event of the execution steps is
never used here (except for the very first "Unpack" step that
processes the initial MBS event directly, without a provider step). There are
no histogramming actions in the execution steps. To view the result data one
has to use a dynamic list histogram or perform a TTree::Draw on the output
event's tree, if existing.
These analysis steps do not perform any
analysis work at all, but only make sure that their own input event is always
set correctly for the following execution steps, depending on the data flow
situation. Generally, there are two cases:
-
the provider step reads the
input event directly from a branch of a ROOT tree (TGo4FileSource). In this case,
the input event remains the native input event of this step as created in the
step factory.
-
the provider step refers to the
result event of a previous execution step.
In this case, the provider processor
itself has to find the correct event pointer by name from the Go4 object management.
The default Go4 framework mechanism to handle these two cases will not suffice
here, since it was designed for a subsequent order of steps and not for a mesh
with parallel execution branches.
To do this job, all provider steps use
the TMeshProviderProc class as general event processor, and the TMeshDummyEvent class as pseudo output event. The TMeshDummyEvent is necessary,
because the Go4 framework will always call the Fill() method of the step's output
event to execute any action of the step. So TMeshDummyEvent::Fill() calls
method TGo4ProviderProc::SetRealInput() to set the pointer to the desired input event correctly.
If the input event is not read from file
(native input event of this step), the provider processor has to search for it
by name using the method TGo4Analysis::GetEventStructure("name"). However, the Go4 framework so far does not offer any additional
parameter to specify the name of the appropriate input for a provider step.
Therefore, this example uses the trick to derive the event name search string
from the name of the provider processor itself: the name of this processor (up
to the "_") is the name of the required event. Note that TGo4StepFactory forbids to use same names for different objects, since the object
name is used as pointer name in the ProcessLine() call; therefore the processor name can not be identical with the
input event name, but must differ by the "_" extension.
Additionally, the provider steps use the
new partial input feature of the TGo4FileSource class (since Go4v2.9). The name
of the event structure defines the name of the TTree branch that should be read
from the input file. The first three provider steps use different parts of the TMeshRawEvent each. If the input event name is set to the name of the corresponding
tree branch (e.g. "RawEvent.fxSub1"), the file source will only read this branch from the tree.
If the input event name is set to the full name of the raw event ("RawEvent",
commented out in this example), the complete event is streamed, including the
not used parts. Note that in both cases the event object must consist in the
full TMeshRawEvent, although in the partial input case only one sub-event is filled.
This is required for a proper event reconstruction due to the ROOT TTree
mechanism. In this example, the partial event input might increase the process
speed by a factor of 2 compared to the full event input.
Although the step configuration can be
defined as usual from the analysis configuration GUI, not all combinations of
enabled and disabled steps make sense to process a subpart of the complete
analysis mesh. For example, if execution step 2 shall be processed, the
corresponding provider step for its input event has to be enabled, too. Note
that the standard step consistency check of the Go4 framework is disabled here
to run such a mesh at all (SetStepChecking(kFALSE)). So it is user responsibility to ensure that all required event objects
are available for a certain setup. Moreover, with >13 analysis steps the standard
analysis configuration GUI becomes quite inconvenient.
Therefore, the example uses a Go4
parameter TMeshParameter for the easy setup of the configuration. This parameter has just a
set of boolean flags to determine which execution step shall be enabled.
Depending on this setup, the UpdateFrom()
method of the parameter also enables or disables
the required provider steps. However, the parameter does not contain the full information
of the input file names for the providers yet (In a "real"
application, this could be implemented in a similar way though).
Thus the configuration procedure looks
like this. The TMeshParameter is edited on the GUI to enable the desired execution steps. The parameter
is send to analysis and switches the steps on and off. Then the analysis
configuration GUI has to be refreshed by the user pressing button to view
the new setup. Here the user may change the names of the event sources for the
provider steps, if necessary. After submitting these settings again from the configuration
GUI, the mesh setup is ready. Note that once the mesh is configured in this
way, the configuration can be stored completely in the analysis preferences and
restored on the next startup.
One could also think of a user defined GUI
that handles both the setup of the TMeshParameter, and the rest of the analysis configuration in one window. This
would offer the additional advantage that it could show the structure of the
analysis mesh in a graphical way. However, such a user GUI is not delivered
here, but can be created according to the hints given in package Go4UserGUI
(see chapter 6.16, page 80).
One way to test the example could look like this:
The following picture shows the GUI with all elements. On the right side you see the Go4 browser. The left side will be the display panel. Below is the Tree viewer, and under this the message window, the mbs monitor, and the analysis status display. With Show/Hide in the Settings one can configure the layout and save/restore it. All buttons in the top row are also available as pull down menus commands.
gui300
This would be the minimal look of a running analysis (the date is updated from the analysis):
There are several screen movies on the Go4 web showing the use of
the Go4 GUI.
There are many keyboard shortcuts to handle windows and actions. See chapter 11, page 92.
The icons in the top line are grouped into
three segments corresponding to the first three pull down menus File, Tools, and Analysis.
Pull
down |
Icon |
Function |
File |
|
Open: opens local ROOT file |
|
|
Open Remote: open TNetFile, TWebFile or TRFIOFile to access remote data |
|
|
Open HServer: open conntection to gsi histogram server |
|
|
Save memory: save content of the memory browser into a ROOT file |
|
|
Close all files: close all ROOT files opened in file browser |
|
|
Exit: closes
window and exit from GUI |
Tools |
|
View Panel: creates window (canvas) to display histogram(s) |
|
|
Fit Panel: opens fit panel |
|
|
Histogram properties: opens window showing histogram properties |
|
|
Create New His: opens
histogram creation window |
|
|
Condition properties: opens window showing conditions properties |
|
|
Condition Editor:
opens central condition editor |
|
|
Event
Printout: examine
current event contents |
|
|
Create Dyn. List Entry: histogramming
on the fly |
|
|
Load Libraries: opens tool to load ROOT libraries |
|
|
User
GUI: starts user GUI |
Analysis |
|
Launch Analysis: starts up the analysis task (as client or server) |
|
|
Connect to Analysis: login to running analysis server |
|
|
Prepare connection: allow external analysis client
connect to this gui |
|
|
Disconnect Analysis: remove connection without analysis server shutdown. |
|
|
Shutdown Analysis server: in administrator mode only! |
|
|
Set+Start: submit setting and start analysis |
|
|
Start: start analysis events loop (after setup and submit) |
|
|
Stop: stop analysis events loop |
|
|
Configuration: open the configuration windows |
|
|
Analysis Window: opens the output window of the analysis |
The help menu provides manuals on-line.
gui357
In the Settings pull down menu as shown on the left side one can set Fonts and Style, Preferences and Panel defaults, where one can set the histogram Statistics box and view panel layout. Preferences specifies when objects are fetched automatically from analysis.
You can adjust all fields according your needs. Then Save Settings. The next start of the GUI will restore the saved layout. Note that settings also contain other preferences, like window geometry and tools visibility, view panel background color and crosshair mode, graphical marker appearance, connection setup parameters, etc. By default, the settings are stored in text files ./.go4/go4localrc and ./.go4/go4toolsrc. To get the standard setup one may delete these two files. If the current directory does not contain a Go4 settings file on Go4 GUI startup, it will be created using the global account preferences at $HOME/.qt, or from the standard installation settings.
Settings
behavior can be changed using environment variable
With the Show/hide entry of the settings menu (or with RMB in an empty menu region) one gets the window on the right to select which tools shall be visible. The actual content of these windows is preserved even if they are not displayed. This is also available as popup menu when clicking the right mouse button on an empty field of the main window.
The Log actions of the GUI can be
defined in a setup window from the settings menu. By default, the log output
(e.g. condition properties, histogram information) is printed into the shell
window where the GUI was started from. Additionally, a text file may be
specified for output. Logging mode specifies if log output is produced On demand only (i.e. on clicking the log button when available), or Automatic whenever the content
of an editor/information window changes. Priority defines the level of output suppression: Errors, Warnings, Infos, or Debugs. Level Errors will only log in case of
an error, Debugs
will printout even debug information of the Go4 kernel. This reflects the priority
of the TGo4Log::Message() method.
In the Panel defaults►Canvas color menu the default background color for all newly opened view panels
can be set. This color may be saved together with the other settings. Panel defaults►Marker labels specifies the default label layout.
The Crosshair mode entry toggles the
default crosshair cursor on/off for all newly opened view panels. This
crosshair state may be saved together with the other settings. However, the
crosshair can be switched independently for each pad in the menu of the view
panel (see chapter 6.7, page 59).
The Generate hotstart entry will save the current state of the GUI (window geometry, objects in memory and monitoring list, objects in view panel, analysis settings) to a Go4 hot start file (*.hotstart). The name of the hot start file can be defined in file dialog here. When re-starting the Go4, the hot start file may be used as command line argument, restoring the state of GUI and analysis (see chapter 6.15, page 80).
With Terminal history the buffer size for the analysis output window can be limited.
The Windows pull down menu shown on the right side provides items to arrange the windows and to save and clear the analysis and log windows.
To access data from user defined classes (like parameters or events) a library including the ROOT dictionary is required. This library is produced by the make file and has the name libGo4UserAnalysis.so. It is recommended to load user libraries for non-Go4 classes (for instance, user event classes) before opening a file with a TTree, where object of these classes are stored. There are three different ways to do it.
First, any external shared library (with or without ROOT dictionary inside) can be loaded by press of the button on the main window. A file dialog then asks to specify the library to be loaded.
Second, set the
environment variable
export
Third, the new possibility (since ROOT 4.00/08) for automatic load of libraries with a .rootmap file. This file contains information to automatically load all necessary libraries for user classes. All make files of the Go4 examples generate .rootmap files during compilation. To explicitly generate this file again, type make map after compilation. If this file is located in the current directory (where GUI is started) or in the user home directory, all libraries will be loaded automatically at the time when required. For more details about .rootmap files see the ROOT home page.
Press the button (or Alt a n or Strg n). This will start the Launch analysis window to execute the analysis task on another host. The operation mode of the analysis task may be “As client” (default), or “As server”; this has to be specified in the selection box on top of the Start Anaylsis window.
The difference of these modes is that in client mode the analysis connects as client to the starting GUI and will be finished when the Go4 GUI terminates. There can be only one GUI connected to an analysis in client mode. The starting GUI will connect automatically to the analysis client after launching it with full controller priviliges.
In contrast to this, the analysis started
“as server” will be an external process independent of the starting GUI. Therefore
in server mode the analysis can not run
embedded into the Qt Window of the GUI.
Any number of Go4 GUIs may connect to this analysis server with
different priviliges, but only one GUI may be the authorized controller. Especially
the starting GUI has to login to the analysis server after launching it in a
separate dialog window.
Besides the selection of the operation mode, the popup window
expects an arbitrary name for the analysis and the node name of the machine where
the client should be started. Normally this is the current node (localhost)
as offered by default. Furthermore there are fields for the user working
directory (in this directory the analysis is started) and the program name. Note
that these values are stored to, and retrieved from the current Go4 settings
file. Start the analysis with button Start
or
RET.
The client will
be started by script AnalysisStart.sh in a remote
shell (Starting mode rsh), or secure shell (Starting mode ssh). One should add
in this script definitions needed by the analysis. The output is directed to a
text window inside the GUI (“Qt Window”), or to an external Xterm, or to the KDE konsole
(if existing), depending on the selected Shell mode.
After
initialization the client connects to the GUI. When this procedure is done, the
message "Starting analysis client ...Please wait” changes to "Editing Analysis Configuration ..." and the GUI is ready popping up an analysis terminal window
and the analysis configuration window. Here the analysis steps can be configured
(see chapter 6.4, page 48). Then the analysis must be set up by pressing Submit (or Alt u).
After setting up
the analysis it is started by (or Alt a s or Strg s). In the browser the directory of the remote Analysis appears. The
next figure shows the GUI with a running analysis. On the right side is the browser
with the analysis directories; on the left side the analysis terminal, and the analysis
configuration window.
gui305
The configuration window is described in more detail in the next chapters.
To
launch the analysis in server mode, the Operation mode in the start dialog window must be switched to “as server”. The
other settings are the same as described in section 6.3.1, except for the disabled possibility to run the
analysis shell in the internal Qt window of the GUI. Immidieately after
starting the analysis server, the “Connect to analysis server” dialog will pop
up, expecting specifications for login of the GUI to the newly created server.
Once the analysis server has been started (from the start dialog, from other GUI, or from external shell command line, respectively), one can connect this GUI to the server. This is done via the “Connect server” dialog that is available from the connect button in the analysis menu. If the server has been launched before from this GUI, the connect dialog will popup automatically.
Port number must match the connection port as printed out in server terminal window. Host should specify the node name of the server machine.
Three different accounts (roles) for login are provided: Observer, Controller, and Administrator. Each login has to be verified by a password. The Go4 default passwords go4view (observer), go4ctrl (controller), and go4super (administrator) are used when the default check box near the Password field is active. They may be changed in the MainUserAnalysis program by methods
TGo4Analysis::SetObserverPassword(),
TGo4Analysis::SetControllerPassword(), and
TGo4Analysis::SetAdministratorPassword(), respectively. In the latter case, the correct password must be typed into the password field.
Only one controller or administrator may be logged in at the analysis server at the same time. If a controller (or administrator) GUI has already been attached, the next controller or administrator login will get an observer role. Observers may only view analysis objects and configuration, but may not modify them. Submit, Start and Stop, and remote macro execution is forbidden for observers, too. The controller account may modify all objects and the analysis setup and change the analysis running state, but may not shutdown the analysis server itself. Finally, only the administrator account may terminate the analysis server. After connection is established, the GUI main window title will show the role ( <Observer> , <CONTROLLER>, <ADMINISTRATOR>).
When a controller after a connection wants to change the configuration he must open the Analysis Configuration window and get the configuration with . When the configuration is OK, submit. When connected as an observer, button may be used to get the object list from the analysis in the browser. You may also get the configuration, but cannot submit.
The GUI disconnects from the analysis by , but the analysis continues. To really shut down the analysis one has to use button (administrator only).
The MainUserAnalysis program must be adjusted to work as analysis server. The constructor of TGo4AnalysisClient got two additional arguments: servermode and autorun. The usage can be seen in Go4Example2Step/MainUserAnalysis.cxx. To run in server mode and controlled from GUI, servermode must be kTRUE. When started from the GUI, the analysis is started by MainUserAnalysis –server name.
In this case servermode is set kTRUE and autorun kFALSE.
Alternatively, the analysis can be started manually in server mode using the batch argument list with the first optional argument –server. This sets both servermode and autorun kTRUE. Then the analysis starts immediately using the setup specified by the arguments. Note that the preferences file is not used! This is useful when one wants to start MainUserAnalysis manually to be connected by GUI clients (on-line).
For analysis servers in ROOT macros see chapter 7, page 82
The Analysis configuration window shows the last valid setup of the
analysis steps. These are taken from the user analysis constructor parameters,
or from the ROOT file Go4AnalysisPrefs.root (in analysis working directory), if existing.
The Analysis
configuration consists of the configuration parameters for each analysis step.
The analysis steps are shown in different tabs of the configuration
window. The values for event source, event store and working status of the
analysis steps can be changed for each step separately. Depending on the chosen
Event Source, different parameter fields will highlight for optional parameters.
The MBS File, e.g., can specify an MBS tag file name (see MBS manual), and numbers for the first event, the last
event and the event number step
between subsequent events to be processed. Multiple input metafiles are
supported by a preceding @ character (see chapter 6.4.5, page 50). The Event Source Remote Event
Server may need a Port number, other on-line sources can set the socket timeout in seconds. For user defined
sources (see chapter 6.4.6, page 50), the optional string argument Args may be evaluated in the
user step factory.
The Event Store settings define the ROOT
split level and branch buffer size of the ROOT tree, and the
file compression level. If the Overwrite radio button is false,
new events will be appended to a previously written tree of the same event
store name.
Moreover, steps
may be disabled completely: the first step, e.g., can be left out and
the second step may read its input from a previously created output file of the
first step. Note: the input of the actual first step must be specified; otherwise
the analysis will not be initialized!
The auto-save
file for analysis objects (histograms, conditions, parameters, dynamic list
connections) is defined for all steps with the auto-save interval, the file compression
level, and the Overwrite
option. Selecting never for the auto-save interval will prevent saving the objects during
the analysis run. However, the auto-save file will be written at the analysis
shutdown (resubmit next settings). Auto-saving can be disabled completely by
the DISABLED checkbox,
i.e. the auto-save file is not even opened for reading previous objects.
Note that the buttons at the different name fields will open
a browser for the local file system to search for appropriate file names.
The new settings
are activated on the analysis client by pressing the Submit button (or Alt u). Note: you have to
press Submit even if you want to apply the settings unchanged! To synchronize the configuration window with
the current analysis settings, the refresh button can be used. This is usually done automatically
on first connection of the analysis, but it might be useful when starting the
analysis manually from a different shell, or when changing the analysis setup
independently from the GUI.
The Submit button closes the previous analysis (i.e. all files and connections
will be closed, all event classes except for the analysis step factories will
be deleted) and initializes the analysis with the new settings.
To have the
changed settings available on the next analysis client startup, press the Save
Button . This
will write the current analysis settings to the file Go4AnalysisPrefs.root
(default name for startup), or to any other ROOT file specified in the file
dialog or the filename text field. Previously written configurations can be
loaded using the Load button and the corresponding file dialog.
·
Note 1: A changed configuration must first be submitted to the
analysis before it can be saved.
·
Note 2: When a new configuration is loaded, the previously active analysis
is closed without saving the configuration. After loading a configuration
it appears in the configuration window. To initialize the analysis with
these new settings, the submit button must be pressed!
When using the Qt Window option in the launch window, the analysis terminal window of the GUI shows all analysis printouts
gui133
Button clears
the window, prints all histograms info, prints all conditions info (make window wide
enough for the counter bars). Button will
open the event information window (see chapter 6.14, page 79).
Additionally, it is possible to kill the analysis process with the button on the hard way. This will disconnect the analysis client after a while from the GUI and analysis can be launched again. However, this is not recommended since the ROOT output files may remain in a non valid state after the kill!
Analysis terminal output can be stored by Windows►Save Analysis window menu command to text file. Analysis terminal output history is limited by 100 Kbytes. This value can be changed in Settings►Terminal history menu command. To keep full history, 0 should be set.
When the analysis task is running in an external shell (xterm, konsole), the buttons and macro execution line will appear in a special dockwindow (see figure).
The analysis terminal window offers the possibility to execute ROOT
CINT commands and macros in the analysis task. Note that a history of previous
commands of the session is available with the macro line combo box (mouse selection,
or arrow down key). looks up for macro files.
Using the go4 pointer (already set to TGo4Analysis::Instance() ), one has access to all public methods of the analysis framework
from inside the macro. Note that the shortcut @ exists here for TGo4Analysis::Instance()->, e.g. @PrintHistograms(“Cr1*”) will print all histograms with names matching the wildcard
expression. In macros the environment variable __
It is not necessary to load the Go4 libraries in the macro again, since these are known at runtime in the analysis anyway.
See also macro
execution in GUI (see 6.17, page 81). $
When auto-save is enabled (in MainUserAnalysis), all objects are saved into a ROOT file after every auto-save interval seconds time, and before termination. The auto-save file can also be written on demand by Save button in the configuration window. At startup of the analysis the following actions are done:
When the analysis is controlled from GUI, objects are loaded from auto-save file when the Submit button is pressed (full sequence see chapter 4.5.9, page 28)
There is the possibility to process multiple input files (source type MbsFile) in one analysis set-up. This can be achieved by wildcard characters in the Event Source name field, e.g. *.lmd or data???_march03.lmd or *. All files matching the wildcard expression will be read subsequently without closing the analysis; output events may be written into one event store. Additionally, one may specify the name of a metafile containing a list of inputs; the metafile name has to be preceded by an @, e.g. @gaussfiles.lml. Each line of the definition file gaussfiles.lml may contain the following format (values separated by blank spaces):
inputfile tagfile firstevent lastevent
skipevents
The numbers of first
and last event always refer to the running event count in the currently open
event source, starting with number 1 each (not the event number inside the
event header). The skip events number defines how many events shall be skipped
in one file in between two processed events; this may be useful if a long term
sample of a large input file shall be taken. The tag file may contain information
which events shall be processed in the input file (see MBS manual).
At least the input file name must be specified; wildcards are not allowed here. Complete lines in the metafile may be commented out by a preceding “!” or “#” character.
Moreover, metafile lines preceded by an @ character are treated as ROOTCINT commands, e.g.
@ .x setup.C
@ TGo4Analysis::Instance()->ShowEvent(“Unpack”); .
These commands are executed in between change of event source, thus allowing to use different setup parameters for different list-mode files.
Note that
multiple input files also work in batch mode.
However, wildcard expressions must be put in parentheses (“”) if they are
passed to the MainUserAnalysis as command line parameter. In batch mode the input file suffix is
automatically expanded to *.lmd, if it was neither .lmd nor .lml. Therefore the meta file can also have suffix .lmd, i.e.
@myfiles results in reading myfiles.lmd (although it is
a plain text file). A better way is to use suffix .lml, because then
one can omit the @ and therefore the parentheses.
Besides the
delivered Go4 event sources for the standard MBS or ROOT file input, there is
the possibility to define any other event source. In the analysis configuration
window, there is the selection UserSource for the
analysis step Event Source type. In this case, a TGo4UserSourceParameter object is passed to the step factory of the step. The user source
name, and optionally, port number and a text argument can be specified in the
configuration GUI to be evaluated on analysis initialization. Method CreateEventSource()
must be re-implemented in the user step factory to
react on a TGo4UserSourceParameter by creating any kind of TGo4EventSource subclass that the user had defined for his purpose. Note that
method CreateInputEvent() should also be overwritten to create a raw event matching to the
user event source, since the default of the base class TGo4EventServerFactory always delivers a TGo4MbsEvent.
The package Go4ExampleUserSource shows a simple example of a user defined event source reading data
from an ASCII text file. Like the two step example, the package can be copied
to a user working environment, and the class names can be renamed replacing the
“TYYY-” prefix.
The event source class TYYYEventSource is prepared to handle any ASCII file containing columns of data separated by blank spaces. Each row is read and its values are converted in order into the Double_t fdData array of the raw event class TYYYRawEvent. The array expands automatically depending on the number of columns. Lines starting with “!” or “#” characters are treated as comments and are ignored. Thus these two classes need not to be modified for input of any ASCII files of that type. However, both the unpack procedure as specified in the event processor TYYYUnpackProc , and the unpack event class TYYYUnpackEvent, are depending on the column’s meanings here and must be adjusted. Additional information can be found in the README.txt file of the example package.
When working with the gsi multi branch system mbs as event source, Go4 offers a monitoring tool that can request information from the status port of a running mbs system. This is available as dockwindow from the “Show/hide” menu, or will appear when the mbs button is pressed in the analysis configuration window.
gui327b
The screenshot shows the Go4 main window with the mbs monitor tool docked in the bottom part, right above the Go4 analysis status line. The mbs monitor by default shows just one line of information, but may be extended by the lower line with more details using the More... checkbox.
The upper line displays, from left to right: The mbs logo which is animated when the mbs acquisition is running; a text line to edit the mbs host name; event rate (events/s); total events acquired; data rate (kB/s); total data acquired (Mb); time and date of last refresh. On the right there are control buttons: With the mbs status server is newly connected and the information is refreshed. It is possible to refresh the status frequently, this is switched on and off with the buttons and , respectively. The refresh time can be chosen by the “seconds” spinbox in the lower line of the mbs monitor window.
Additionally, the lower line displays (from left to right): Name of data server in use (streamserver, or eventserver) and percentage of delivered events 1/n, as it is set in the mbs by command set stream n, or set event n; percentage of real delivered events from this data server; name of the file which is currently written by the mbs, if existing, and total amount of data written to file since mbs startup. The button may be used to print the complete mbs status structure, the complete setup structure, or the multilayer setup structure, respectively, to the shell from which the gui was started. This is selected by the radiobuttons Status, Setup, and SetupML. Note that printout of multilayer setup is enabled only if a real multilayer setup exists in the observed mbs.
Besides the time selector for the monitoring frequency, the right side of the second line offers the possibility to switch on several trending histograms. This is done by the trend checkbox. The overall number of bins may be changed in the bins selector; the range of one histogram bin equals the monitoring frequency. Note that trending histograms are only written if the mbs status monitoring is turned on (i.e. no new entry in trend histogram by manual refresh using button ). Three different trending histograms are currently produced: for the event rate, the data rate, and the percentage of delivered events at the mbs data server (streamserver or eventserver). They appear in the Go4 browser in the Workspace/Mbs folder and may be observered in Go4 view panels. The screenshot shows the trending histograms for event rate and streamserver event ratio. Note that a warning sign will appear in the upper line if connection to mbs status server fails.
When working with the new gsi DAQ framework
DABC (Data Acquisition Backbone Core, see
http://dabc.gsi.de) as event source,
Go4 offers a monitoring tool that can request info services from the DIM server
(Distributed Information Management, see http://dim.web.cern.ch/dim/)
of the DABC control system. It is available as dockwindow from the “Settings/
Show/hide” menu. Note: this tool is
available only if Go4 installation was build with environemt variable $DIMDIR
set to the DIM installation directory!
gui362
The picture shows the Go4 main window with the DABC monitor tool at the bottom and some histograms filled with monitored data. On the left side of the monitor tool (frame DIM dns server) the hostname which runs the DIM name server can be specified. When pressing the button, the list of DIM services is refreshed from the dns server. The time and date of the most recent refresh of any requested DIMinfo appears in the text line below the dns server name. The more... checkbox allows to show or hide the Inspector frame that contains more controls (see below).
Any node that exports a DABC state service will show up in the DABC Nodes list (table in the middle); the first column gives the XDAQ Context name in the form nodename:port, as specified in DABC setup. The current value of the DABC finite state machine (or the MBS acquisition status, resp.) is marked with its corresponding colour and name in the State column of the nodes table, e.g. “Running”.
By using the Show checkboxes in the Ratemeters column the user can select to subscribe for all DABC rate parameters of the corresponding node. These subscribed rate parameters will show up in the Ratemeter View list (table on the right). The first two columns of the Ratemeter View list denote the full DIM parameter Name (unique for each DIM name server scope) and the current Value (in the units as published by the DIM service, here Mb/s). The table information is refreshed by means of the DIM mechanism whenever a rate service is updated. Note that a warning sign and the number “-1” will appear in the Value column if a previously known service variable is currently not available. This may happen when DABC is switched into “Halted” state which will remove DABC modules and their rate parameters.
The histogram checkboxes in the Trending and Statistics columns of the Ratemeter
View table select which ratemeter values shall be filled into a trending or
statistics histogram, resp. These histograms can be found in the “Workspace/DABC”
folder of the Go4 browser, sorted into subfolders according their full context
names. They can be displayed in a viewpanel, renamed, saved to a file, etc.
like any other Go4 histogram (See chapters 6.5 and 6.7).
The histogramming is controlled by the Inspector frame (on the left). The histogramming on checkbox toggles the histogramming of the selected values on or off. Before turning on the histogramming, the user may chose the histogram binsize by means of the binsize spinbox. Note: when binsize is changed, all previous trending/statistics histograms will be replaced and lose their contents! If each sample is checked, there will be fast trending (or statistics, resp.) histograms which are filled at every update of the DIM service (usually in 1 s interval). If average is checked, there will be average trending (or statistics, resp.) histograms which are updated with the average of the values accumulated during a user defined time interval. This time interval is set by the spinbox next to the average checkbox (here 5 s). Both each sample and average histograms may be accumulated simultaneously; if neither each sample nor average is checked, there will be no histogramming.
Besides observing the rate parameter values, it is possible to print the current values of any DIM service exported on the DABC nodes. The Dump checkbox in the Service list column of the DABC nodes table selects a node for a full service list inspection. The button in the Inspector frame may then be used to print the current values of all selected services to stdout of the shell from which the gui was started. Moreover, it is possible to dump services only if their name contains a certain text, or if it matches a shell-like wildcard expression. This filter expression may be written into the text field near the button before pressing it. In the screenshot example shown in the figure, only parameters with names containing the string “Rate” would be dumped.
After pressing the analysis starts and the rates are
displayed at the bottom as shown in the screen shot below. The analysis output
window and the configuration window have been closed. A view panel created by has been opened and a histogram is displayed
by dragging & dropping a histogram from the browser into the canvas. Note
the logging window displaying messages from the remote analysis. This log panel
can be opened in the Settings menu bar. The complete
logging history may be saved into a text file by the Windows►Save Logwindow menu command.
gui309
The Go4 browser on the right side shows objects from different data locations in a folder structure. Remote objects in the connected analysis task are listed under the Analysis branch. The Workspace folder contains all objects that are put into the memory of the local GUI, e.g. by creating fix copies of remote analysis objects. A root file opened from the files toolbar with the button will appear in a folder of the filename; similarly, a connection to a remote data source like the xrootd, the root webfile, or the gsi histogram server, shows up as separate browser branch.
Beside the “names” column showing the objects in their folder structure by symbols, the Go4 browser has configurable columns to display different kinds of properties of the displayed objects: Flags, Info, Date, Time, Class, and Size. These can be switched on and off by the menu that pops up on right mouse button click in one of these. Moreover, the order of these columns can be freely arranged in the browser by dragging and dropping their caption to a new position.
The Flags column will indicate certain properties of the object by letters:
· m - shall be monitored frequently; or s - is static until explicitely refreshed
· d - object may be deleted; or p - is protected against deletion
· r – read only, can not be reset; or w – writable, may be reset
The Info field will usually show the type of the folder, or the title of the ROOT object.
Date and Time columns show the date or time of the last object refresh to the GUI inernal cache (for remote data sources), or of the object creation (for local workspace), respectively.
Class column shows the class name, and Size will give an overall object size in bytes.
gui311
Each item in the browser has a context menu, which can be activated by right mouse button click on that item. It is shown in the figure on the right. By means of this menu, it is possible to operate on the browsed objects. The items in the upper part of the context menu (above the line) are available for all items, whereas the items in the lower part contain special functionality to control remote data sources like the analysis.
Histograms and pictures can be plotted either by double click, or by drag and drop in a view panel, or by the right mouse menu. Item Plot draws each selected histogram into an own graphical pad, Superimpose draws all selected histograms superimposed on one pad.
The browser items represent the structure of a connected data source like the remote analyis, but will only retrieve the objects on demand. This happens usually just before the objects are drawn. To explicitely get the objects into the local memory cache without drawing them, the Fetch item(s) functionality may be used. Note that the browser’s implicit fetching behaviour may be adjusted in the Settings/Preferences menu by “Fetch when drawing”, and “Fetch when copying”.
The selected objects may be saved into a ROOT file with menu item Save selected... The Export to... functionality will offer the possibility to export root histograms to ascii or radware format.
Item Info shows some information of the object, Edit... opens the editor if available. Item Delete Item deletes the selected objects from the local memory, whereas item Delete from analysis will delete the corresponding object in the remote analysis, if possible (see chapter 6.5.8. page 57).
The Analysis folder shows the remote
folder structure, which contains all objects that were registered to the
analysis client. At any time the list of the remote objects may be refreshed by
the right mouse button entry Refresh nameslist. The folder Histograms e.g. contains the histograms, the folder Trees will show the structure of all registered trees, e.g. all trees created
by TGo4FileStores.
The eraser item Clear (Reset to 0) clears the selected objects like histograms, conditions, graphs and so on.
Each object on the
analysis has two protection modes – delete protection and clear protection.
These modes indicated in Flags column of analysis browser (see below). Delete protection is set
for an object when it is created and added on the analysis side. It prevents
deletion of such objects from GUI. Objects created by GUI commands have no such
protection and can be deleted by the Delete from analysis functionality. Clear protection prevents the user to clear the
content of objects by using . This
mode can be set and unset for any object via context menu commands Set clear protection and Unset clear protection
, respectively.
In the Analysis a histogram, graph, or picture can be set into the monitoring mode by selecting it and pressing the monitoring entry Monitor item(s) in the right mouse menu. This is indicated by the letter “m” in the Flags column of the browser (static objects have letter “s”). Monitoring means that the content of objects are updated continuously from the remote data source (analysis, histogram server,..) to the GUI. This allows e.g. to watch the filling process of a histogram. The monitoring property of an item may be switched off by the Stop items monitoring functionality of the context menu.
Note that only the visible objects are frequently updated, i.e. even if a browser object is in monitoring state, it will not be copied from the remote data source if is not drawn in any viewpanel, or displayed in an editor, respectively.
The overall monitoring action can be started with button of the Browser options dockwindow. Here the update frequency may be specified in seconds,too. Button will cease monitoring of all monitored objects, but will not change their monitoring property (flags). Additionally, this dockwindow offers a button for immediate refresh of all visible objects, and a filter function for the browser to display either all objects, or only the monitored objects, or only the currently fetched objects, respectively.
The Workspace folder contains all objects that are put into the memory of the
local GUI. This may happen either by producing a new histogram from the ROOT
menus in the viewpanel, like a rebinning, or a projection, or from the Go4 tree
viewer; or objects may be copied from elsewhere to the workspace. Item Copy to Workspace will produce a copy of the current object and put it into the
workspace folder. This copy will preserve the subfolder structure of the data
source; if e.g. a histogram was copied from analysis folder “Histograms/Crate1”,
the copy will be placed in folder “Workspace/Analysis/Histograms/Crate1”. The Copy to clipboard, Paste from clipboard, respectively, allow a standard copy/paste functionality to any
destination in the workspace. Additionally, in the workspace folder the right
mouse button menu offers the Create folder and
the Rename object functionality, as known from general file system browsers.
ROOT files
containing data can be opened (buttons and of the mainwindow file menu, respectively) as
with the native ROOT TBrowser/TTreeViewer. Any ROOT file can be opened. Histograms in these files can be displayed
in the Go4 view panel like local objects. A ROOT tree in a local file can be
examined with the tree viewer of Go4. In contrast to the remote tree viewer
mode, trees in a local file are processed by the GUI itself and do not have an
effect on the remote analysis. The GUI knows if a tree viewer entry comes from
a remote, or from a local TTree, so the button will either send a command to the
analysis client for a dynamic histogram, or will perform a local TTree::Draw() call.
If the file contains user objects, make sure
that the GUI has loaded the proper libraries to access them (see chapter 6.2, page 45).
From the main window File menu entry one can connect to any GSI histogram server
like MBS, GOOSY, LeA, or another Go4 analysis. The parameters for the histogram
server, such as node name (Server), login name (Base), the socket Port number, the Password, and an optional Filter expression, are specified in a
connection dialogue window. After a successful connection the histograms of the
server appear in the Go4 browser in a folder named HServ_basename, if
basename is the name of the histogram server base.
Any object in
the workspace may be deleted by selecting it and using the popup menu delete item . Objects in the Analysis (histograms, conditions, parameters, …) that
were created in analysis code must not be deleted, for the
compiled user analysis would still try to access these objects after deletion.
Therefore, deleting these objects is disabled using the delete protection
property (symbol “p” in Flags browser). However, dynamic objects that
had been created from the gui (histograms, conditions, dynamic list connections)
are not delete protected and can be removed by the delete button.
An analysis
histogram can be reset (contents and statistic values to zero) by selecting it
and pressing the clear button except Clear is disabled. Resetting an
analysis TGraph object will erase all points of the curve. For parameters, the
method Clear() is
called which may be implemented by the user. All objects within an analysis
folder are reset at once by selecting the folder icon in the remote browser and
pressing button . This
has the same effect as calling method ClearObjects(“Foldername”) of TGo4Analysis. Note that any analysis object can be protected against clearing by
a switch in the remote browser’s right mouse button context menu (See chapter 6.5.3).
The Go4 tree
viewer is started via Settings►Show/Hide►Tree viewer menu or via RMB pull down menu.
gui140
There are two operation modes for the Go4 tree viewer: the local mode, or the remote mode. Dragging and dropping the tree leaf names from file or remote browser, the tree viewer will switch automatically into the local or remote mode, respectively.
The tree viewer works on a tree in a file that was opened in the browser. This is like the original ROOT tree viewer, with the same logic of drag and drop. However, the Go4 tree viewer supports the resolution of the Go4 composite event information (see section 7, page 82). On pressing button , the local tree will be processed as defined by the given draw expressions in (and optional ) fields of the Go4 tree viewer. The local histogram of the given name is filled with the result. The histogram will appear in the memory tab and may be displayed in a view panel. If no name is specified, an automatic name is chosen from the given leaf names.
All classes, which are stored in the tree, should be known to GUI. User should load appropriate libraries before using local tree viewer (see chapter 6.2, page 45).
The Analysis folder shows the
structure of all objects registered to analysis trees in the Trees subfolder.
By drag and drop the elements of a tree can be put into fields of the Go4 tree viewer. A name and an
optional drawing condition can also be defined here. The logic is the same as
for the regular ROOT tree viewer. On pressing button , this
information is passed to the analysis client and a new entry in the Go4 dynamic
list is created. After pressing in the Analysis panel, a new histogram of the
defined name appears in the histogram folder (if no name was defined in the
tree viewer, a default name is used combining the variable names). Note: the
histogram itself will be created no sooner than the next events after the are processed, i.e. the analysis must be
running. This histogram will be filled event by event with the defined
parameters of the tree. Go4 internally uses a TTree::Draw() over a number of
collected events to update the histogram contents. This number, the dynamic
list interval TreeDrawInterval, can be set by the analysis method SetDynListInterval(Ndyn), or can
be changed in the dynamic list editor (see chapter 6.12, page 76).
If the histogram
specified in the tree viewer already exists when the dynamic list entry is
created, the histogram of that name will be filled by the dynamic list instead
of filling a new histogram. Therefore it is possible to create a histogram with
desired bin size first (see chapter 6.6.3, page 58), and then assign this histogram to a new entry of
the dynamic list. This can be done easily by dragging and dropping a histogram
icon from the histograms folder into the histogram textbox of the tree viewer. Again,
pressing will create the dynamic list entry; the given
histogram will then be filled every Ndyn events. The dynamic list tree is kept in memory, if in the analysis
configuration for output Go4BackStore had been selected.
A histogram
filled by the dynamic list, like any other remote histogram, can be displayed
continuously in a view panel by switching on the Go4 monitoring mode (see chapter
6.5.4,
page 56).
The button will popup the histogram creation window. Here
the properties of the histogram to be created anew can be specified
(dimensions, precision, binning, range, name, title). The histogram may be either
created in the local directory (Create Local), or created in the remote analysis (Create
Remote). A new local histogram will appear in the
local objects panel, a remote histogram is put under the histograms folder in
the Go4 folder structure. A new histogram (like any existing histogram) can be
used as target for the remote or local tree viewer. This is done by specifying
the histogram name in the tree viewer name field, or by dragging and dropping
the histogram icon to this name field. The tree viewer will then fill the created histogram instead
of creating a new histogram with arbitrary binning and range settings.
Pressing in the Go4 main control window opens a new Go4
view panel. A new view panel will also pop up automatically when any object in
the browser is selected and the right mouse button menus or are activated. Furthermore, objects can be
drawn by “drag and drop” from the Go4 Browser to an existing view panel pad and
displayed there. On the left side the optional ROOT graphical editor is embedded.
It is opened by Edit►Show ROOT attributes editor. Select with left
mouse an object on the canvas and the editor will change accordingly.
gui318
gui368
The view-panel
offers the menus:
Save as.. save the
content of the view-panel in different formats.
Print ... hardcopy the view-panel to
$PRINTER or .ps file
Produce
Picture create Go4 picture from
viewpanel, put it in workspace
Close the view-panel
Show Marker Editor open marker panel
Show ROOT Attributes open ROOT graphics editor
Show Event Status toggle ROOT event status in bottom
line
1:1 coordinate ratio adjust pad margins to 1:1
coordinate ratio
Default pad margins restore default pad margins
Clear Markers clear all
marker objects in pad
Clear Pad clear contents of current pad (and sub-pads)
Clear Canvas removes content and pad divisions
When
histograms or graphs are displayed in
superimpose mode, each one may be selected here. Then attributes like color may
be set for selected histogram. If the selected object is currently not on front
of all superimposed objects, an additional menu entry “Show ... on top”
will appear. When chosen, this entry will pop the selected object to the
foreground. Note that the object first
must be selected and then set to top.
Crosshair toggle the ROOT pad crosshair mode
Super Impose toggle superimpose option
Histogram Statistics toggle display statistics box on pad
Multiplot Legend show legend for superimposed
histograms
Histogram Title toggle display histogram title on pad
Draw Time display refresh time in histogram title box
Draw Date display refresh date in
histogram title box
Draw item name display full path and name in histogram title box
Keep View panel Title Do not overwrite title
Set View panel Title Set the title
With Settings->Panel defaults one
can set defaults for these values. If the Superimpose option is selected, any new histogram that is dragged to this pad
will not replace the existing histogram, but will be displayed in the same pad
with the old one (as ROOT THStack). A legend box will show the graphical style and the name for each
drawn curve. This legend can be toggled on or off with the Multiplot
Legend option. The text of each legend entry can be
changed by opening the right mouse button popup menu at the entry position and
using the SetEntryLabel function (see ROOT TLegend
class for documentation of further methods in this menu).
An existing view panel can be divided into independent sub-pads by the division buttons in the Canvas Tools activated with the RMB on an empty region. When several histograms in the browser are selected for plotting, the view panel division will be done automatically to display all histograms in one new view panel window. Graphic style and range settings are always applied to the sub-pad that was selected most recently (red frame which is set by middle mouse button in ROOT), except the Apply to all option checkbox is enabled.
It is possible to extend the regular histogram title by information on the refresh time and date by switching on the options Draw Time , and Draw Date ,respectively. Additionally, the full name of the displayed object, i.e. the complete path and item name in the Go4 browser, may be displayed in the histogram title by toggling the Draw item name option.
Usually, the
title of the view panel window (showing up in the Windows menu of the main Go4 window) is taken from the object that was
drawn most recently in one of the sub-pads. This behavior can be changed by
options Keep View panel Title and Set
View panel Title , respectively. This allows to
specify a meaningful name for a view of several histograms that will not change
when one histogram is exchanged by drag and drop on a sub-pad.
The Show
event Status option in the edit menu will display the current mouse
coordinates and histogram channel contents in the bottom line of the view
panel. If the canvas is divided, this information always refers to the selected
pad.
The canvas
embedded in the Go4 View panel is an ordinary ROOT canvas, offering all ROOT features
of the mouse button actions on the displayed objects (e.g. opening a histogram
fit panel, rescaling the axes using cursor and left mouse button, ...). The
active pad must be selected with middle mouse (ROOT). After using ROOT popup
windows the active pad must be selected again! Note that the settings are
preserved for each pad!
The view panel
may be saved to a file in several formats by choosing File►Save as.
The buttons (gui141) are zoom and shift buttons for the x-, y and z- axes, working on the active pad. In multi pad view panels the active pad must be selected with middle mouse (ROOT, red frame). After using ROOT popup windows the active pad must be selected again! The expansion/compression factor can be set in % of the current range. The Un-zoom all button will restore the complete range of all axes. The set limits button will popup a scale window. Here the range can be typed in and set explicitly by axis values. Additionally, the scaling behavior of the ROOT histogram can be changed: By default (AutoScale on), the y-axis (1D histogram) or z-axis (2D histograms), respectively, is expanded to cover the full range of channel contents whenever a memory histogram is updated, or when a monitored histogram is refreshed from the analysis. With AutoScale disabled, the previous y-range (1D) or z-range (2D), respectively, is invariant over any updates. This allows to observe a magnified region of interest in a spectrum, independent of the maximum peak height. Note that the y range of a 1D histogram can be chosen freely by ROOT TAxis selection with the mouse, i.e. clicking with left mouse button on the y-axis for the first limit, and dragging the pressed mouse to the second limit of the range. The scale window is automatically connected to the selected pad and updated accordingly.
The draw options can be set by two menu bars: One for all options available (Settings->Show/hide->Draw Options) steered by pull down menus and one for a subset (Settings->Show/hide->Draw Options Short) steered by buttons only:
gui366
The draw options follow the ROOT draw options (see next page).
gui367
Go4 option |
Description |
ROOT |
scatter |
black scattered points |
HIST |
pixel c |
colored pixels |
|
cont c |
colored contour |
CONT |
surf c |
colored surface |
SURF2 |
pix+scale c |
colored pixels and color scale bar
|
COLZ |
cont+scale c |
colored contour and color scale bar |
CONTZ |
Gouraud |
smooth grey scale surface |
SURF4 |
lego c |
colored lego |
LEGO2 |
lego/shadow |
lego with one side colored |
LEGO1 |
lego bw |
black and white lego |
LEGO3 |
mesh c |
colored meshed surface |
SURF1 |
mesh bw |
black and white meshed surface
|
SURF |
mesh+cont |
bw meshed surface and colored contour on top |
SURF3 |
line c |
colored contour lines |
CONT1 |
line dot bw |
black dotted contour lines |
CONT2 |
line bw |
black contour lines |
CONT3 |
boxes bw |
black boxes |
BOX |
digits bw |
channel content as numbers |
TEXT |
ASImage |
TH2 as TASImage (fast pixel map with scale bar) |
|
P0 (1D) |
Polymarker without lines |
P0 |
L (1D) |
Line |
L |
C (1D) |
Smooth curve |
C |
B (1D) |
Bar chart |
B |
mesh+cont2 |
bw meshed surface and colored contour on top |
SURF5 |
cont4 |
colored contour |
CONT4 |
cont1+ pal |
colored contour lines and color scale bar |
CONT1 |
cont4+pal |
colored contour and color scale bar |
CONT4 |
arr (2d) |
arrow plot |
ARR |
In a view panel a marker panel can be opened by Edit►Show Marker Editor menu item:
gui324
Pressing once on button and then one more time in the pad, a channel marker (cross) with a label and a connecting line is drawn. Once created, any marker can be re- positioned by chosing its name in the marker selection box and using again the button: the next pad click moves the currently active marker to the picked position. If new is chosen in the marker selection box, a new marker is created and added to the list. Note that the selected marker is always displayed on front of all other objects in the pad. Clicking on a marker or its label box with the left mouse button will also pop it frontmost.
With new selected and loop option enabled, the cursor stays after in point marker mode. Subsequent clicks in the
pad create new markers. This behavior also applies for the other marker types,
respectively:
draws a window marker (with two subsequent LMB clicks) and a label.
draws a polygon marker (TCutG): each click will define one point of the polygon, a double click
will finish the definition of the shape.
X: places a (Latex
formatted) label.
draws an arrow from first click to second click.
In loop mode one can switch between the five marker types.
outputs the values of the markers to the activated log output.
A selected markers can be deleted by pressing the button near the marker selection box. Furthermore, markers may be deleted and configured with RMB on the cross or inside the window, respectively (see right TGo4Marker menu: DeleteMarker and
left TGo4WinCondView menu: DeleteRegion). The setter methods configure the layout through little windows as shown above (options 0 or 1, then apply and cancel). All elements can be moved with LMB (labels are updated). SaveLabelStyle applies current settings to all subsequent markers. With Settings►Save settings in the main Go4 window menu these settings will be stored. With Edit►Clear Markers one can remove all marker elements. To change the graphical attributes one can use the new ROOT graphical editor. It should be opened by Edit►Show ROOT attributes editor. When a graphical object is selected (LMB) the editor changes accordingly. Close the editors also through the Edit menu.
With Settings►Panel defaults►Marker labels one gets the window shown below. Here the default layout can be
specified and saved.
A condition may be displayed in an existing viewpanel by dragging and dropping it from the browser to a destination pad containing a appropriate histogram. The full condition editor (see 6.8.2) may also draw its working condition to the viewpanel.
It is possible to edit any condition displayed in a viewpanel already by means of the marker editor in the bottom line (see figure).
gui330
Condition wincon 1 is drawn above the histogram Sum1 that is filled only if this condition is true. As the regular markers (see 6.7.6), the condition may be selected by name in the marker selection box. In addition to the control buttons for the markers, editing a condition will enable some more buttons in the marker editor. After changing the condition by moving its boundaries, a will appear to remind you to update the condition by button on the analysis side. With the current condition state from the analysis side is refreshed in the editor window. If working on a condition from file, the refresh button will appear instead to reload the viewpanel condition.
Button opens the info window for the selected condition (see 6.13) to view current conditon properties that are not displayed in the viewpanel label. For advanced editing of the condition, the full condition editor may be invoked using button .
The condition editor window is popped up when one double clicks on a condition in the browser or using the edit function of the browser’s right mouse menu. It may also open by using the in the viewpanel marker editor.
In addition to the features of the marker editor, it may display and change all properties of the Go4 condition class, e.g. counters, testing properties, histogram statistics over the region, etc.
The following figures give some examples.
gui328
Window condition cHis1 displayed with histogram His1. The histogram has been bound to the condition by method SetHistogram() in the analysis. In this case the histogram is automatically displayed when the condition is edited.
Polygon
condition polyconar is a polygon condition
array from the two step example which can be displayed in a 2d view panel. When
a condition array compound is
edited, the index of the currently active condition can be set in the upper
right spin box. The displayed values always refer to the selected array member.
When selecting an entire condition array in the editor (All button
or spin box index “-1”), changes will be applied to all members.
gui329
With the button the active pad of the current view-panel (selected with middle mouse button) is set as display working pad for the condition. The condition is drawn on this pad until the display button is pressed again with another active pad. If the working pad contains a histogram, it is assigned to the condition under edit and its name is shown in the editor. Note that it is possible to exchange the condition work histogram by drag and drop of a new histogram into the condition editor display pad.
After editing
the condition limits graphically on the working pad, the changes will be
updated automatically whenever the mouse enters the editor window. When a
condition is changed in the editor (always press Enter to confirm changes), the
graphical representation will be updated automatically. After changing the
condition, a will appear to remind you to update the
condition by on the analysis side. With the current values
(e.g. counters) from the analysis side are updated in the editor window.
Conditions can be set to return always true or false, respectively. The result
of a condition check can be inverted. A polygon condition checks, if a point
(x,y) is inside a polygon (TCutG). A window condition checks, if one or two values are inside one or
two intervals, respectively.
A condition has counters for the number of all Test() calls performed, and for the number of true results. The counter values after the last refresh are displayed in the editor. With these values are reset to zero and the condition is directly updated on the analysis side.
The button allows to pick the boundaries of the
condition region with the mouse. This works in the same way as in the marker
editor: for window condtions, two subsequent clicks will take the click
position as limits (for 2d conditions, these clicks define corner points); for polygon conditons, each click will set a
corner point until the mouse double click finishes the pick mode.
Button outputs the current condition values to the GUI starting window, or into a log file if specified in the Settings menu (see 6.1). Button saves the condition in a file. If the condition editor is working on a condition in a ROOT file (via File Browser), the button will update the changes in the original file by default. This is useful to edit conditions in an existing auto save file.
The condition
editor offers four tabs: for the condition limits, for the display properties,
for the statistics inside the selected condition range, and for the mean
values, respectivly They are shown in
the next screen shots:
The Limits tab contains the values of the window condition limits, or the largest extension of the polygon condition boundaries. These are updated from the graphical representation on the working pad, or can be typed in directly in case of window conditions (to apply the typed values press RETURN).
The Cut tab is only active for polygon conditions. It shows the table of x and y coordinates of the polygon (TCutG). These values may be edited here (to apply the typed values press RETURN). Moreover, the number of polygon points can be changed with the NPoints selector box. If the TCutG is edited graphically on the pad by mouse, the values in the table will be synchronized the next time the mouse enters the editor window.
The Draw tab shows the names of the histogram and viewpad used to display the
edited condition, and allows to control some draw properties. Each condition
can be set as visible or not with the visible checkbox. If visible, the condition is shown on the working pad,
otherwise it is hidden. This is useful when working with condition arrays. It
is recommended for polygon conditions to improve editing. The visibility is a
property of the condition class itself and is stored in the auto-save file. The
label checkbox
enables the drawing of a graphical label together with the condition (see screenshot
examples). This label may contain the limits values from the Limits tab; this
can be toggled using the limits checkbox. Other entries of the label may be configured in the Stats and Mean tabs.
The Stats tab shows some statistics
(Integral, position and channel content of the maximum) of the current histogram
inside the selected condition. In addition, the Mean tab contains mean and RMS values for x
and y directions. Setting the corresponding checkboxes plots these values into
the label on the working pad.
In the next
example two conditions are bound to the upper pads of a picture (see chapter 6.9, page 72) by method AddCondition().
gui335
The histograms in the lower pads are filled under the condition shown in the pad above. All picture conditions will be shown simultaneuosly (if their visible property is true). Mouse click on a picture’s subpad will deliver the names of all contained conditions into the selection box of the marker editor. The selected condition may be modified and updated by means of the marker editor, or using the full editor started by button, as described above. The mechanism to bind conditions to picture pads guarantees that a condition is set always on the correct histogram.
With the button of the main window “Tools” menu and toolbar, one can open a window to create a new condition in the analysis. This functionality is available as a shortcut from the dynamic list editor, too (see 6.12). The Create condition dialogue expects a condition name, the type (1-D window, 2-D window, polygon), and optional an array size. For Array size “no array”, a single condition is created, otherwise a condition array compound that contains the given number of conditions.
After pressing the Create remote button, the new condition will appear in the subfolder Analysis/Conditions of the Go4 browser. The name field in the create dialog may contain any subfolder path relative to this default location, e.g. Name: myconditions/region2 will create new condition region2 in folder Analysis/Conditions/myconditions. Non exisiting subfolders are created in this procedure together with the conditon.
Once created,
the condition can be modified from the condition editor or from the viewpanel
marker editor as described above. When
the auto-save mechanism was enabled, the condition will be restored at next
analysis startup. Note that it’s not possible to create a new condition without
the analysis connected to the gui!
The TGo4Picture class provides a way
to set up a view in the analysis, which then can be displayed in the Go4 GUI. A
picture contains:
The following code creates a simple
picture, which contains only one histogram:
TGo4Picture* pic = new
TGo4Picture(“pic1”,”picture title”);
pic->AddH1(histo); // histo is variable of type TH1*
A picture can be
divided into sub-pictures like a ROOT canvas can be divided into sub-pads. The
division of a picture can be specified in the picture constructor or by method SetDivision(int
ndivy, int ndivx) which creates ndixy*ndivx
sub-pictures inside the picture. Sub-pictures can be accessed via method Pic(posx, posy). For each picture (and sub-picture) one can specify the following
options:
Display header pic->SetDrawHeader()
X axis range pic->SetRangeX(double, double)
Y axis range pic->SetRangeY(double, double)
X log scale pic->SetLogScale(0, bool)
Y log scale pic->SetLogScale(1, bool)
Z log scale pic->SetLogScale(2, bool)
To add an object to be drawn the following
methods can be used:
TH1, TH2, TH3 pic->AddH1(TH1*)
THStack pic->AddHStack(THStack*)
TGraph pic->AddGraph(TGraph*)
TGo4Condition pic->AddCondition(TGo4Condition*)
Each method
requires a pointer to the correspondent object and optional draw options (if
necessary). When an object has been added to a picture, the following drawing
options can be set for this object (see ROOT manuals):
Line attributes pic->SetLineAtt(Color_t,
Style_t, Width_t)
Fill attributes pic->SetFillAtt(Color_t,
Style_t)
Marker attributes pic->SetMarkerAtt(Color_t,
Size_t, Style_t)
Draw options pic->SetDrawOption(Option_t *)
TStyle attributes pic->SetStyle(TStyle*)
Axis rebining pic->SetRebinX(Int_t ngroupx),
pic->SetRebinY(Int_t ngroupy)
For example, to configure a picture with four sub-pads (2 x 2), each
with a different histogram, the following code can be used (first index top down, second left right):
TGo4Picture* pic = new
TGo4Picture(“pic1”, ”picture title”, 2, 2);
pic->SetDrawHeader(kTRUE); //
displays time, name and title of picture
pic->Pic(0,0)->AddH1(histo1);
pic->Pic(0,0)->SetRangeX(100,
200);
pic->AddH1(0, 1, histo2); // or pic->Pic(0,1)->AddH1(histo2);
pic->Pic(0,1)->SetDrawOption(“lego”);
pic->AddH1(1, 0, histo3, ”lego”);
pic->AddH1(1, 1, histo4);
AddPicture(pic); // add picture to
frame work
Similarly the colors in next figure have been set up by:
Color_t his=0;
for(int i=0;i<8;i++) for(int k=0;k<8;k++){
fPict1->Pic(i,k)->SetFillAtt(his,1001);
fPict1->Pic(i,k)->SetLineAtt(his,1,1);
his+=2;
}
The TGo4Picture class supports arbitrary levels of picture divisions. This means
that each sub-picture can also be divided. For instance, a picture with 3
histograms, two in top row and third in bottom row, will be created by the
following code:
TGo4Picture* pic = new
TGo4Picture("pic","pic title",2,1);
pic->SetDrawHeader();
pic->Pic(0,0)->SetDivision(1,2); // divide top widget on two more pads
pic->Pic(0,0)->Pic(0,0)->AddH1(histo1); // add histogram to sub-sub-pad
pic->Pic(0,0)->Pic(0,1)->AddH1(histo2); // add histogram to sub-sub-pad
pic->Pic(1,0)->AddH1(histo1,
"lego2"); // add histogram to
sub-pad
AddPicture(pic);
gui337
Current limitations of pictures are:
In the Go4 GUI
pictures will appear in the analysis browser in the Pictures subfolder. Together with the picture all correspondent histograms
will be automatically transferred. Double click on a picture draws it in a new
view panel. A picture also can directly drag-and-dropped into an existing view
panel.
Pictures also
can be put to the monitoring list. Putting a picture to the monitoring list
automatically puts all histograms of the picture to the monitoring list, too.
All information
of a fit like models (= fit functions)
and their parameters, references to the data, and the results are stored in a fitter
object (=FO). The
fit panel (activated by button) is the editor of fitter objects. The
fit panel is attached to a fitter object to edit it. Fitter objects are
stored in two different locations:
To create
fitter for active pad, Fitter►create for pad menu item or
Use pad button of fit panel should be used. The fitter
object can always be copied to memory browser and than saved to the file. The
data reference of a fit object is changed or set when:
§
creating or copying a fitter
object to a pad,
§
dragging a histogram into a pad
(the fitter object of the pad gets the reference to that histogram),
§
dragging a histogram name into
fit panel.
The next picture shows a pad in a view
panel and the fit panel. The peak finder tab is shown.
gui129
On the bottom of fit panel there are five
buttons:
Use pad If fitter displayed in fit
panel, it will be copied to selected pad in last active view panel. If there is
no fitter in fit panel, a new fitter will be created for this pad.
Find Executes peak finder
routine. All peak finder parameters should be setup first. Work only in Wizard
mode.
Fit Executes fit.
Draw Draw models,
backgrounds and model components as sets up in Settings sub-menu.
Pars Show all fitter
parameters in a table. Parameters can be listed one by one or in lines mode, when one line corresponds
to one model and contains amplitude, line position and line width.
There are three different layouts of fit
panel, which can be chosen in Tools sub-menus:
Simple Contains several buttons to fit data to polynomial function,
gaussian, lorentz and exponent.
Wizard Intuitive and easy-to-use tool to setup data objects and model
components. Also includes peak finder setup. Suitable for most fitting tasks.
Expert Advanced
tool, which gives full control over the fitter. Provides a hierarchy view of
all objects inside fitter and possibility to change any relevant data fields.
Supports all functionality, which may not be presented in Wizard tool.
In wizard mode
there are three different peak finders available (see previous figure). Variant
2 is ROOT, Variant 1 searches peaks having specified width range above a
threshold, variant 3 searches minima and maxima using a dynamic noise bandwidth.
Variant 3 also allows for summing up channels to reduce the noise. Depending on
the histogram characteristics, either of these may give good results. One has
to play with the parameters. Changing parameters automatically launches a Find.
Found peaks
are marked in the View panel pad in red. One can move their position and change
their width with the mouse. Clicking on a data or model entry the right side of
the panel shows related information. Models can be [de]activated clicking on
the OK boxes
or removed by [-]. New
models can be added by [+]. After the fit the results can be seen pressing the Pars button (which changes to Back to switch the view back):
gui130
Fitter sub-menu
has following items:
Create for pad create appropriate fitter for selected pad in last active preview
panel
Delete delete fitter
Save to browser save fitter to Go4 memory browser
Update reference updates references on data objects from file or memory browsers
Print parameters produces parameters printout, parameters page should be active
Rollback parameters restore value of parameters, which automatically stored before last
fit
Close close fit panel
Settings sub-menu contains following items:
Confirmation For each
delete action (of fitter, data, model and so on) confirmation message will appear
Show primitives Show graphical
primitives for model position and width and for range settings
Freeze mode Fit panel is
not automatically attached to selected pad, but only by create/copy/move command
from Fitter sub-menu
Use current range At any fit or peak
finder action automatically uses range which is currently selected on histogram
Save with objects Save objects, to which
fitter have references, together with fitter. When such a fitter will be
loaded, it will have copy of saved objects. Available only in expert mode
Draw model Draw model
of data
Draw background Draw background (sum of
all model components, belongs to background group)
Draw components Draw all model
components, which are not belong to background group
Draw on same pad Use same pad for drawing
or create separate preview panel
Draw info on pad Draw on pad info box
with parameters values
No integral Do not
show any integral values on parameters page
Counts In
lines mode on parameter page additionally shows counts number for every model
component inside specified range
Integral Shows
integral value for every model component inside specified range
Gauss integral Calculates and
shows theoretical (based on amplitude and width parameters) integral for
one-dimensional gaussian components. None of specified range conditions are
taken into account.
Recalc gauss width For gauss components
recalculates sigma values to full width on half maximum (FWHM)
Do not use buffers Do not use any memory
buffers for fit
Only for data Use buffers
only for data objects
For data and models Use buffers for all data
objects and model components
Individual settings Use buffers as selected
individually for each data object and model component
Detailed help on fitter and fit panel can be obtained from the main window Help►Fit tutorial.
Parameters are objects containing a user defined structure of
values. These can be applied for controlling and calibrating the user analysis
apart from the analysis framework configuration. All user parameters should be
subclasses of TGo4Parameter. They can be created in the user analysis code and are registered
to the Go4 framework by method AddParameter(TGo4Parameter*
mypar). Once a parameter was registered, it appears
in the Go4 Parameters folder, it is saved and can be restored from the auto-save file,
and it can be edited and updated from the Go4GUI by means of the parameter
editor.
Note that the Go4 GUI has to load the libGo4UserAnalysis.so to edit and save any user defined parameter object from the analysis.. See 6.2 how to load libraries to GUI.
Double clicking
a parameter icon in the browser will open the parameter editor as seen in the picture. All known members of
the user parameter class and its base classes are shown here with their names,
their type and their current value.
gui125
Currently supported types are:
all basic signed and unsigned types, e.g. Double_t
fdEnergy; Bool_t fbIsOK;
the ROOT TString class to wrap text strings, e.g. TString
fxMyFilename;
pointers to TGo4Fitter objects, e.g. TGo4Fitter*
fxUnpackfitter;
and arrays of these in 1 or 2 dimensions, e.g. UInt_t
fuVal[42]; Float_t ffVoltage[5][100];
Comments behind member declarations are
shown in the Comments
column.
Aggregations and pointers to basic types are not supported at the
moment (except for aggregated fitter objects).
Arrays of data
are expanded and collapsed in the table by double clicking on the array name.
Additionally, the right mouse button will open a popup menu to navigate through
the array without expanding it completely.
The values of the data can be edited after double clicking in the value field of the data member table. Note that any editing action has to be finished by pressing “return”, “tab”, or “cursor” before it is valid. To apply the changes, press which will update the edited parameter on the analysis side. This is done by method UpdateFrom(pointer to new) provided by the user class. This means that arbitrary functions can be executed! The changing of data members is fully controlled by the user class. Vice versa, will refresh the table shown in the editor from the current values of the analysis parameter. Note that all changes not yet applied to the analysis or saved are overwritten on refresh!
If one is working on a parameter loaded from a file, button will appear instead of , doing a refresh from the source file. Note that the original parameter in the file is not changed by the editor immediately; the root file is updated only when using the save button . Then a save dialog window will appear, that allows either overwriting the original parameter, or saving the changed object to another file.
Finally, will erase all editable fields of the table. will close the editor without modifying the analysis parameter.
Sometimes it
might be useful to exchange a Go4 fitter object between the analysis and the
GUI. A fitter, e.g., may be prepared using the FitGUI and then sent to the
analysis client where it can be applied to some histograms during analysis.
Vice versa, one might want to display the resulting parameters of automatic
fits in the analysis on the GUI. Therefore, the Go4 parameter concept supports
the TGo4Fitter class as aggregation member, i.e. a pointer to a fitter can be accessed
by means of the parameter editor.
The Go4
framework already offers the parameter class TGo4FitterEnvelope that contains one fitter object. This fitter may be accessed in the
analysis by method GetFitter(). In this case it is important that the fitter object itself is
exchanged inside the parameter each time the parameter is updated. Thus the
user should not keep the pointer to the fitter in his/her analysis class, but
request the fitter from the (persistent) TGo4FitterEnvelope parameter
with the getter method when the fitter should be used.
Additionally, any
user defined subclass of TGo4Parameter may contain references to several fitters or even arrays of fitter
references. Here it is the user responsibility how the fitters refresh their
settings in the UpdateFrom() method. Moreover, one may implement getter and setter methods for
the most important values of the fitters without the need to access the internal
fitters directly. An example is TXXXCalibPar in the Go4Example2Step directory.
Pressing the right
mouse button over the name of a fitter member will open a context menu.
Selecting Edit... will open the Go4 FitGUI
window (see chapter 6.10, page 72). A copy of that fitter is put into the local workspace
of the Fit GUI to be edited or to be applied on any histogram. Selecting Update
fitter from fitgui workspace
in the context menu, the fitter in the parameter
object is replaced by a copy of the fitter that is currently active for the Fit
GUI. So any fitter existing on the GUI may take the place of any fitter inside
a parameter. Note that the original fitter member in the parameter will be lost
after this action unless it is refreshed by from analysis again! To send the changes in
the fitter back to the analysis client, like for all parameters the button must be pressed.
Note that in
case of a fitter pointer array (e.g. TGo4Fitter* fxFitters[10] ),
the context menu will show both the items to manipulate the array view and to
edit or update the selected fitter.
The Go4 dynamic list is a mechanism to connect the event data
with a histogram and a condition. The histogram is filled from certain data
members of the event during the analysis.
Optionally, the histogram may be filled only if a condition that is
tested against other data members of the event is true. In contrast to the
histograms filled from the compiled user analysis code, the dynamic list offers
the possibility to define these relations on-line during the running analysis.
The dynamic list and all newly created histograms and conditions may be stored
in the Go4 auto-save file and are recovered on the next analysis initialization ( or Submit button in the configuration
menu).
In the Go4 browser, the dynamic list folder
contains all existing dynamic lists (currently only one default list). Each list
shows the existing dynamic entries by name. Double clicking on a dynamic entry
will open the dynamic list editor to display and change it.
To create a new dynamic entry, button of the main window tools menu will open the create new entry dialog window. Here you can define the name and the kind of the dynamic entry. There are 2 different kinds of Go4 dynamic entries: The TreeEntry and the PointerEntry (see below). After pressing “Create remote” button, the new dynamic entry will appear in the browser in analysis subfolder DynamicLists.
To delete a
dynamic entry completely, select its icon in the Go4 browser and select in the right mouse button menu.
gui339
Depending on the
kind of the entry, different sub-pads of the editor are enabled: The Histogram and TreeDraw sub-pad for the TreeEntry, and the Event
data and Condition sub-pad for the PointerEntry, respectively.
Any dynamic
entry can be enabled or disabled by switching the enabled checkbox. A disabled entry will not be
processed, but is still in the dynamic list. Note that if a dynamic entry fails
on initialization (e.g. unknown object names), it is disabled automatically.
To apply the changes, press which will update/create the edited entry on the analysis side, respectively. Vice versa, will refresh the values shown in the editor from the current status of the analysis dynamic entry. Note that all changes not yet applied to the analysis are overwritten on refresh! A label will appear near the update button if the changes have not been applied to the analysis yet.
If one is
working on a dynamic entry loaded from a file, button will appear instead of , doing
a refresh from the source file. Note that the original dynamic entry in the
file is not changed by the editor immediately; the root file is updated only
when using the save button . Then
a save dialog window will appear, that allows either overwriting the original
parameter, or saving the changed object to another file.
will clear the target histogram in the analysis to zero counts, and will
reset the events in the backstore tree (in case of tree draw entry, see below.)
This allows to observe changes of the
dynamic entry setups directly if the target histogram is monitored. will close the editor without modifying the entry.
The editor offers the additional feature to
get some information of the histogram and condition status from the analysis.
Clicking in the Histogram or in the Condition sub-frames will retrieve and
display the current object status in the histogram or condition status windows,
respectively (see chapter 6.13, page 79). This may be useful to check if histogram or
condition settings (dimension, ranges, bin size, etc.) are suitable, without
requesting these objects in the browser. Additionally, some filling and testing
statistics is shown here. The GUI tool tips show brief explanations for each
information line.
The button prints the names and connections
of all existing dynamic entries to the analysis output window.
New histograms or conditions may be created
in the analysis by the or the button, respectively. For histograms, the standard histogram creation
window (see chapter 6.6.3, page 58) pops up. Use the Create
Remote button
here. For conditions, the “new condition” dialog is started (see chapter 6.8.5, page 69).
Go4 uses the ROOT TTree::Draw() mechanism for the on-line
evaluation of the data. This works just as described in the ROOT users Guide: A
string expression defines which leafs of the tree shall be scanned by name. Additionally,
the name of the output histogram must be specified; the histogram may either
already exist (Create Remote from Go4 ), or it is created from the first TTree::Draw() by ROOT with automatic range and
binning. Instead of a Go4 condition, this mode works with a TCut string expression to filter the histogram filling.
Note that the TTree must exist for this mechanism.
Usually, the TGo4FileStore output will create and register
a tree that can be used here. If no file output is needed, one can switch on
the TGo4BackStore output (configuration window)
which will fill a temporary TTree in memory that is cleared after
each TTree::Draw() scan of the dynamic list. The TTree::Draw() is not performed for each single
event, but after a number of events have been filled into the tree. This number can be specified in
the user analysis by TGo4Analysis::SetDynListInterval(Int_t val) or by
the Interval
field.
A new tree draw entry can be created either
from the Go4 tree viewer (drag of the tree name from the Analysis browser and
press ), or from the Create Dynamic Entry dialog. In the latter case, the tree name, the histogram name, the draw expression
and optionally a cut expression may be specified directly in the dynamic list
editor after creation. This works by “drag and drop” of historams and tree
leafs from the browser to the corresponding fields of the dynamic list editor.
Note that the TTree name is recognized automatically from the
dropped leaf.
The advantage of a tree draw entry is that
it can access any level of substructures of the event if it is resolved in the TTree (depending on split level); the Go4 composite event data may be fully
accessible here. It offers all functionality of the ROOT TTree::Draw(). The disadvantage is that you
need to fill the event data into a tree to access it. The histograms are not
filled event by event, but the tree is processed in event buffers. The buffer
size should be adjusted by the user depending on the typical event rate. Since
the pointers to the data and the histogram are searched by name for each Draw() call, the performance is slow compared
to histogram filling from direct pointer access like in the precompiled user
analysis case.
gui340
In this mode (PointerEntry), the pointers to histogram,
event data and an optional Go4 condition are looked up by name once on
initialization of the dynamic list. During the analysis, these pointers are
used directly to test the condition and fill the histogram event-by-event. The
information to locate the pointers is taken from the ROOT TClass information of the user event classes; it is not necessary to fill the
event into a TTree.
For the pointer entry, at least the name of
an existing histogram and one dimension of the event data must be specified.
This is done in the Event data tab of the editor. Usually, for a
new pointer entry the histogram should be created by (see above). The new histogram item must then be dropped from the
browser to the dynamic list window.
The event data is defined by the event name
and the name of the data member of the corresponding event class, separated by
a slash (“/”). The Go4 browser Analysis folder offers a view of all existing TGo4EventElements in the EventObjects.Events folder. From here you may just
drag and drop the Data member item to the corresponding field of the
dynamic list editor. Note that data arrays are shown with their maximum size
here, you need to edit the index afterwards to specify the desired array
member.
Similarly, the data to test the condition
can be defined in the Condition tab of the editor. The condition
is usually created and registered in the compiled user analysis and is
identified by name here. Polygon conditions and 2 dimensional window conditions
need the event data specifications both in x and y directions. Note that the
condition event data lines should be left blank if the condition shall not be
tested in the dynamic entry (i.e. the
histogram is filled anyway). With a new condition can be created. Button will open the editor for the specified condition.
The advantage of the pointer entry is that
you do not need a TTree. Testing and filling is done for
each event by pointer without any additional string compare after
initialization. Therefore it is faster than the tree draw. The disadvantage is
that currently only one level of substructures and only one dimensional arrays
are supported (to be improved...). Implicit summing up of not specified array indices,
like in the TTree::Draw(), is not possible here.
To check the properties of a histogram or
condition, general property windows exist for these objects. They support drag
and drop of icons from Go4 browser. These windows will also pop up from the browser’s
context menu when the button
is chosen.
gui341
With the button or the
button of the tools menu one opens the
histogram or condition information window, respectively. To see the properties
of a histogram or condition, drag the icon from the browser into the window.
With the information is updated from analysis. With the information is output to the GUI start up
window, or into a log file if specified in the log settings (see chapter 6.1.2, page 43). With all histograms ( all
conditions) are listed in the analysis output window. starts the condition editor for a condition. displays the histogram in a view panel.
The event information tool window allows to control printout of event samples from the analysis. The button of the tools menu will open the event information window. This button is also available as a shortcut in the Qt analysis terminal. The entry of the browser’s context menu (right mouse button) over an event itemwill open the event information tool, too.
The name of the examined
event is shown in the top text line. By default, the MBS event is chosen for
printout. The event object names may be dragged and dropped to the event information
window from the Go4 browser. Clicking the button will switch to the MBS event mode directly
without the need to drag the MbsEvent-10-1 icon.
The ShowRemote checkbox selects if
the printout of the event sample is done in the remote analysis terminal, or in
the terminal where the GUI was started. The TTreeSample checkbox selects if
the PrintEvent() method of the event
shall be called (TTreeSample off), or if the
sample event shall be written to a ROOT Tree which will use the TTree::Show() method to scan and display the data (TTreeSample on). Note that for user event classes that do not implement a PrintEvent() nothing will be displayed except for the TTreeSample mode.
Each
click on button will print events as shown in the upper part
of the screen shot left side. The examine button will display a new printout of the currently active
event (lower output on the left). Note the different format!
Additionally, for MBS events this window provides in the MBS Event sub-panel parameters for the SetPrintEvent() method. One can specify in the left field how many MBS events
arriving shall be printed out in a special format. In the next field a
sub-event id may be filtered (default is to display all sub-events). The hex checkbox selects to print the sub-event data either in hex or in
decimal format, while the long checkbox defines if the
data is seen as longwords or words. Pressing the button will resubmit these settings to the
analysis thus initiating a new printout of n events. Note that the MBS Event sub-panel is independent of the settings for the regular printout
of the current event. It works for the remote analysis terminal only, and it
uses a different printout format than the TGo4MbsEvent::PrintEvent() or TTree::Show() methods.
When starting
the GUI several actions have to be done to get the analysis running. If these
actions are always the same it would be convenient to save them in a file and
execute this file when starting the GUI next time. This mechanism is called hot
start. The typical actions are:
§
Launch analysis client
After GUI and analysis are configured, one can create a hot start file by Settings►Generate hotstart. A file selection menu pops up were one can specify a file name. The postfix should be .hotstart. The next time one can start the GUI with this filename as argument (.hotstart can be omitted). Then all actions stored in the file are executed.
With care, this file could
even be edited.
Go4 provides a
possibility to execute user widgets on GUI side. There is an example of a user GUI,
included in the standard Go4 distribution in directory $
The easiest way
to create a user GUI is to copy the content of the standard example to another
directory (e.g. ~/UserGUI) and compile it there (make clean, make all). The user should also specify the path to this directory in
export
The
The user can
freely modify any widgets in the example and create new ones. Changes in
library names or the top widget class
should be reflected in the
There is a support of “old style” user GUI, created with older version of Go4 (up to v2.8). In that case correct path to libraries should be specified like:
export LD_LIBRARY_PATH=~/OldUserGUI/Go4Library:$LD_LIBRARY_PATH
gui351
The command line window can be enabled with the settings pull down
menu like all other windows. With button the following menu appears with a list of provided
macros. These can be executed directly or can be used as templates for other
macros. An environment variable __
corrhistos.C takes the bin contents of two histograms and creates a two-dimensional graph.
hishisto.C creates a histogram and makes a distribution of the bin contents of the source histogram.
An interface is provided to access objects from the Go4 browser. A description can be found in the reference manual.
Caution!
Macros running inside the GUI can crash the GUI!
$
The Go4 analysis server offers the possibility to observe and control execution of normal ROOT macros from the Go4 GUI. This allows the development of analysis code without respect of Go4 analysis framework classes (like TGo4EventProcessor, TGo4AnalysisStep and so on) still providing remote access to the running environment of a user analysis.
It is possible with minimal effort to observe histograms, produced and filled by practically any running ROOT script. The script go4Init.C initializes Go4 and starts the analysis server in background. Function go4RegisterAll() then scans the current directory for existing histograms and makes them available remotely.
Usage:
1.
To enable ROOT to find the go4
macros one should enter in the .rootrc a line
Unix.*.Root.MacroPath: .:$(ROOTSYS)/macros:$(
(Note that .rootrc may be in current directory or in $HOME.
The standard provided by ROOT is in $ROOTSYS/etc/system.rootrc)
2.
Run normal ROOT session.
Execute go4Init.C script by command:
root [0] .x go4Init.C
3.
Run user script:
root [1] .x userScript.C
4.
When go4Init() is executed, go4 will start the server and printout the port number
for connection:
"Waiting for client connection on
PORT: 5000"
5. Start the Go4 GUI in and connect to the analysis server running in the CINT. See section 6.3.2 page 46 for more.
The Go4 framework can be accessed after go4Init by the global method
TGo4Analysis* go4=
TGo4Analysis::Instance();
After this call, variable go4 can access any method of the analysis framework.
Any object to be seen remotely by the GUI must be registered by one of the following methods:
·
go4->AddHistogram(his); // makes
histogram TH1* his available in the Go4 GUI
· go4->AddAnalysisCondition(conny);// dito for TGo4Conditions
· go4->AddParameter(par); // dito for TGo4Parameters
· go4->AddPicture(pic); // dito for TGo4Pictures
· go4->AddTree(mytree); // register TTree, but do not change Tree ownership to Go4
·
go4->RemoveTree(mytree); //
unregister TTree: important to cleanup reference in Go4 if tree
//
is removed from ROOT (closing TFile !)
· Please see Go4 Reference Manual for other available Add... methods!
The go4RegisterAll() function (from Go4Init.C) registers all histograms found in the current directory. Some more information can be found in the example macros (see below).
· Int_t seconds=go4->WaitForStart(); Polls until the Go4 is set into the "running" state (by Start button on GUI or SetRunning() method) with 1 second interval. Returns number of seconds from begin of wait until "running" is switched true. If negative value is returned, a ROOT interrupt has happened during wait (e.g. Ctrl-C on CINT Canvas).
· Int_t state=go4->Process(); Process one main cycle of Go4 event loop from macro. Will first execute any command from GUI, second call the Go4 main cycle to process analysis steps, user event function and dynamic list (if existing). This call is required inside any explicit loop in the macro to process go4 framework analysis actions. The GUI event rate meter is also updated by this method. Return value is <0 if running state is stopped, otherwise 0.
·
go4->SetRunning(Bool_t on); Switch Go4 running state from inside a macro. Useful
to react on analysis conditions
· Bool_t on=go4->IsRunning(); Check the running state of the Go4. Maybe obsolete since this is done implicitly in methods WaitForStart() and Process(). However, macro loop may be controlled from GUI independent of Go4 main loop processing.
The following examples can be found in $
·
hsimple.C This is a standard ROOT example from
$ROOTSYS/tutorials. The only modification is to call
go4RegisterAll() after creating histograms.
To run this example, start a regular ROOT session, init the Go4 server and
execute script:
root [0] .x go4Init.C
root [1] .x hsimple.C
· hsimplego4.C A variation of hsimple example. This macro will wait until the Go4 start button is pressed and then run the random filling in infinite loop (mind your disk space, since a TNtuple is filled into a file here!) Registered objects may be monitored. The loop can be started and stopped at any time from the Go4 GUI. Please try the remote tree draw on the TNtuple from the Go4 GUI and view the newly created histograms. Try to launch the TBrowser before executing the macro and inspect the content of the "Go4" folders locally...
·
treedrawgo4.C Macro works on tree in a file. As before,
first execute .x go4Init.C:
root [0] .x go4Init.C
root [1] .x treedrawgo4.C("filename")
The "filename" specifies a ROOT
file "filename.root" that
contains a TTree. Note:
first tree found in file will be used.
This macro contains 2 examples on trees:
Control
of remote analysis macro from Go4 GUI
gui359
Besides the full featured Qt GUI, the Go4 analysis may be controlled and observed by a regular ROOT CINT session, using the native ROOT GUI for display.
The following screenshot shows at the bottom a go4 CINT analysis server task. Here example hsimplego4.C is running (see 7.3). This process is connected with the ROOT session in the upper part of the picture, which uses the regular ROOT GUI to browse and display the analysis objects. This is just like it would be possible with the usual Go4 GUI. Actually, a multithreaded Go4 master task is running in the background of the upper ROOT session, while a Go4 slave task is working on the analysis in the lower root session. This analysis process may not only be a root session with Go4 analysis server, but may as well be a compiled Go4 analysis client executable (MainUserAnalysis).
gui360
The controlling Go4 master process is realized in the ROOT session by the TGo4Interface class. After starting a regular ROOT and loading the Go4 libraries, the call
root [0] new TGo4Interface
will instantiate the master task framework. Explicit loading of libraries is
not necessary if the corresponding ROOT mapfile mechanism is used. Once
initialized, the variable go4
is defined as a pointer to the interface instance and
may use all methods of class TGo4Interface.
Note that in the
analysis server session as described in section 7, variable go4 refers to the class TGo4Analysis instead!
To connect to an existing analysis server,use
root [1] go4->ConnectAnalysis("localhost",5000,0,
"XXXview");
Arguments are: hostname of the server, the port number, the login account (0=observer, 1=controller, 2=administrator), and the password. If password is left out, the default password of this account is used.
Alternatively, an analysis client may be
started from this session using
root [1]
go4->LaunchAnalysis("test","/u/user1/go4",
"MainUserAnalysis","lxi003");
With arguments: arbitrary name (“test”), path to the analysis executable, name of the analysis executable, and node where analysis process shall be started.
The above methods correspond to the Connect analysis and Start analysis dialogues of the Go4 GUI (section 6.3).
Once the connection to the analysis process is established, it can be controlled by several methods:
·
go4-> SubmitAnalysisConfig(); Submit
the analysis configuration. This corresponds to the Submit button of the Go4 GUI. Usually, the configuration is retrieved from
analysis after connection. It may be modified by several methods of the
TGo4Interface before submit, or it may be submitted unchanged. A submit is
required in any case before analysis can be started. Note that this command is
not allowed when logged in as observer.
· go4-> StartAnalysis(); Start the analysis run. This corresponds to the Start button of the Go4 GUI. Note that this is not allowed when logged in as observer.
· go4-> StopAnalysis(); Stop the analysis run. This corresponds to the Stop button of the Go4 GUI. Note that this is not allowed when logged in as observer.
· go4-> StartMonitoring(Int_t period=10); Start monitoring all objects that are set to monitoring state and drawn. The update period can be specified in seconds.
· go4-> StopMonitoring(); Stop monitoring all objects. Will not reset the monitoring property of the objects.
· go4->DisconnectAnalysis(); Remove connection to remote analysis process.
For a complete reference of available
methods, please see the header file $
In addition to the analysis control by
TGo4Interface calls, the regular ROOT browser will offer some extensions after
the connection has been established. Start the browser with:
root [2] TBrowser br; .
If connected to the analysis, there is a Go4 folder among the regular ROOT folders. This will browse the structure of the remote analysis with subfolders and all objects. Both histograms and Go4 pictures may be drawn to a new canvas by double clicking on the item. Go4 conditions will be drawn on double-click only together with the histogram that was bound to it. The ROOT right mouse button menu has entries added for the remote Go4 objects:
· Draw Item - will draw it if possible, just like double-click
· Copy to Workspace - Produce fix copy to the Workspace folder in local memory. Just like in the regular Go4 GUI.
· Delete Item - remove object from analysis if possible
· Set Monitor On/Set Monitor Off - Switch the monitoring property of the selected object
· Toggle monitoring – Start and stop monitoring in general. A dialog will appear to request the monitoring periods in seconds. For zero period, monitoring will be stopped. This corresponds to TGo4Interface methods StartMonitoring() and StopMonitoring().
The status line at the TBrowser bottom will
show the analysis rate meter, and eventually some messages retrieved from the
analysis. Additionally, the status messages are printed out to the CINT terminal.
A "real"
example of the composite event usage can be found on the Go4 web.
Thinking about
possible structure for typical experimental data, one naturally comes to
something more or less close to the following scheme:
·
A so-called event model should be defined as the unit
for the data processing. It represents the record of all physical interactions
in the detector after the reaction between a beam particle and a target. This
event can be real, calibrated or simulated as one process real, calibrated or
simulated data. The event can contain the original data from the different
detectors (Raw level) but also reconstructed data resulting from the different
reconstruction steps ( Calibration, Hits, Tracks ...) As it contains the
complete set of physical data, the event is a complex object. To avoid having a
monolithic event containing everything on the same level, one can take
advantage of structuring it into smaller independent components (sub events).
hangs are easier to perform, extensions are easier to integrate and your design
is much easier to understand. Separating the whole into parts is also an advantage
if one considers using partial IO functionality: one should be able to retrieve
only some part of the whole event from the file i.e. the relevant part for his
analysis, the rest being deactivated in the IO.
·
The decomposition of the event should follow fixed
rules. For example, one can decide to have the scheme as shown in the picture
·
In this example the top-level event is composed by a
global header containing for example the general setup of a typical experiment
i.e. RunID, beam energy, magnetic field value, target type and size etc .... On
the same depth-level a detector folder contains the different detector’s
specific data i.e. Raw , Cal, Hit,... data-levels and eventually a specific
detector header.
·
It should be possible to access directly a part in the
tree-like structure of an event. For that purpose each component should be
uniquely identified.
·
The event is obviously a persistent object. Therefore
the data should be organized into a ROOT tree with a branching layout which fit
to the natural branching of the tree-like event structure.
The design of the
event object should be general enough in order to be adapted by different
experiments, should not create performance penalty compare to native ROOT tree
structure when streaming the objects in file and should provide an easy-to-use user interface hiding the
underlying ROOT internal data representation.
According to the general description, it is
clear that an event should be defined as a composite object. To implement such
a composite object, the composite design pattern has been used. The composite design pattern is an abstraction model that allows a client to
treat individual objects and composition of objects uniformly. It allows to
build complex objects by recursively composing similar objects in a tree-like
manner. It also allows the objects in the tree to be manipulated in a
consistent manner, by requiring all of the objects in the tree to have a common
super-class or interface.
Using such a
general design, one can build complex structure out of simpler one. And then
build even more complex things based on what you have already build and so on
... But you would like to treat all of them the same way.
The following picture shows the UML
diagram of the composite event.
The core classes
are shown in color. An abstract component TGo4EventElement declares the interface for object to be composed. It also
implements default behavior common to all classes.
An abstract composite class TGo4CompositeEvent is defined as a subclass of TGo4EventElement interface and has no specific behavior
itself. The TGo4CompositeEvent acts as a container class of classes inherited from TGo4EventElement and as it is itself deriving from the
same abstract interface, it could contains also classes inherited from TGo4CompositeEvent . Therefore one as no limitation in
composing complex objects in a tree-like structure. Furthermore, this abstract composite class is the natural place for the implementation of pure recursive
algorithms. For instance, when the function TGo4CompositeEvent::getEventElement(
const char* name) is
invoked, it is simply sequentially sent to all its children in order to find
the one matching the name. The composite event class implements child-related
operations.
The user can add
its own components for modeling its own event structure by sub-classing either
directly the TGo4EventElement class or the TGo4CompositeEvent class.
The diagram above shows 2 different user-defined subclasses i.e. TGo4EventHeader and TGo4DataStore which
both contains specific data-members or data-containers that has to be streamed
in a binary ROOT file format. Go4 provides an equivalent of TGo4DataStore: TGo4ClonesElement.
The process of
modeling an event structure will now be described. In order to benefit from the
composite model functionality, one will have to create either sub-classes of TGo4EventElement or
sub-classes of TGo4CompositeEvent according
to the following scheme:
·
Elementary data objects inherit from TGo4EventElement . These objects are elementary bricks of the data structure which contain
members of all data-types that the ROOT system supports in its IO split mechanism.
{ enum (kSize=10);
Char_t fType[20]; // array
of characters
Int_t fNTracks; // single
type int
Int_t fX[kSize]; // an array where dimension is an
enum
Float_t fMatrix[4][5]; // 2-
dimensional array of floats
Float_t *fDistance;
//[fNTracks];
TString fNames[12]; // array
of TString
EventHeader fEvtHdr; // example
of class not derived from TObject
TClonesArray *fData; // array
of cloned TObjects
TH1F* h1; //->
pointer to histogramms
TArrayF fArrayF; // an
array of floats
TArrayI *fArrayI; // a
pointer to an array of integer
}
All these
example data-members can be used when sub-classing the base class TGo4EventElement . The
user-defined class will then be translated into a TBranchElement object that represents the basic element of the ROOT TTree object. The corresponding TBranchElement in turn will contain a list of TLeaf objects for every data member appearing
in the header class definition.
Class MyDerivedClass: public
TGo4EventElement{
//
private data menbers
Int_t data[10][20];
MySpecificObject fMyobj; // specific data-object should not
derived from Tobject
// etc ...
MyDerivedClass();
MyDerivedClass(const Text_t *aName,const
Text_t *aTitle,Short_t aBaseCat);
~MyDerivedClass();
// some public user-defined function
for this class
// etc ...
}
In order to create an elementary object, one should
follow the TGo4EventElement general interface, and according to this interface the data object
should have as parameter of its constructor
·
a name
·
a title
·
a unique identifier
The name will be used to
generate the corresponding TBranchElement branch names in the ROOT TTree layout. The identifier should be unique for each user class. It is good style to define all identifier in a
global header file in the project as follow:
// file MyIndexesDefinition.h
{
#ifndef __ElementIndexes_H
#define __ElementIndexes_H
// Fast Bus nodes position in
composite structure
const Int_t fastbusID=0; // defined reserved
area of index for fastbus object i.e [0,10]
const Int_t Ifastbus1=fastbusID+1;
const Int_t Ifastbus2=fastBusID+2;
const Int_t Ifastbus3=fastBusID+3;
// etc …
// slot nodes position in composite
structure
const Int_t slotID=10 ; // defined reserved
area of index for slot objects i.e [10,40]
const Int_t Islot1 =
slotID+1 ;
…
#endif
}
One can also define
data-container class that inherits from TGo4CompositeEvent. i.e.
// file MyCompositeEvent.h
#ifndef MyCompositeEvent_H
#define MyCompositeEvent_H
#include
"Go4Event/TGo4CompositeEvent.h"
#include
"Go4Event/TGo4ClonesElement.h"
#include "FB.h"
#include "CAM.h"
#include "VME.h"
#include "MyIndexesDefinition.h"
class MyCompositeEvent : public
TGo4CompositeEvent
{
public:
Int_t EvNumber; //
just event number
Int_t FBNumber; //
number of FB data for event# EvNumber
Int_t CAMNumber; //
number of
Int_t VMENumber; //
number of VME data for event# EvNumber
public:
MyCompositeEvent();
MyCompositeEvent(const char* name, const char* title,int index);
~MyCompositeEvent();
void SetT4pEventHeader(Int_t inumber){EvNumber = inumber;}
Int_t GetEvNumber() {return EvNumber;}
void AddF(Int_t FNumber, Int_t Slot, Int_t Chan, Int_t Data);
Int_t Fill();
ClassDef(MyCompositeEvent,1)
};
#endif // !MYCOMPOSITEEVENT_H
The class MyCompositeEvent will create the complex structure of the user-defined
event inside its constructor i.e:
// file MyCompositeEvent.cxx
MyCompositeEvent::MyCompositeEvent(const
char* name, const char* title):
TGo4CompositeEvent(name,title)
{
// create a Composite structure for each
FastBus
TGo4CompositeEvent *fb1 = new
TGo4CompositeEvent("FB1","FastBus1",Ifastbus1);
TGo4CompositeEvent *fb2 = new
TGo4CompositeEvent("FB2","FastBus2",Ifastbus2);
TGo4CompositeEvent *fb3 = new
TGo4CompositeEvent("FB3","FastBus3",Ifastbus3);
MyDerivedClass *data1,*data2,*data3;
data1 = new
MyDerivedClass("DataFB1","TDataFB1",Islot1);
data2 = new
MyderivedClass("DataFB2","TDataFB3",Islot2);
data3 = new
MyDerivedClass("DataFB3","TDataFB3",Islot3);
// add data object to each fastbus composite event
fb1->addEventElement( data1 );
fb2->addEventElement( data2 );
fb3->addEventElement( data3 );
// add One DataObject to each fastbus node in MyCompositeEvent
addEventElement( fb1 );
addEventElement( fb2 );
addEventElement( fb3 );
}
This example shows how to
create an event structure out of elementary user-defined data-objects class
i.e. MyDerivedClass simply using the function :
TGo4CompositeEvent::addEventElement( TGo4EventElement * evt );
The structure can then be browsed
from the Go4 tree viewer.
File pad: open local ROOT file on disk |
|||
|
File pad: open remote ROOT file
(TNetFile, TWebFile, TRFIOFile) |
||
|
Save content of memory to ROOT file |
||
|
File pad: close selected ROOT file |
||
|
File pad: close all ROOT files |
||
|
Export selected objects of memory browser
into another format (ASCII, radware, ROOT) |
||
|
Stop running analysis, shutdown analysis
and terminate GUI |
||
|
Open view panel |
||
|
Open fitter window |
||
|
Open histogram properties window (there:
list properties in analysis window) |
||
|
Open histogram creator window |
||
|
Open condition properties window (there:
list properties in analysis window) |
||
|
Open condition editor |
||
|
Open event inspector window |
||
|
Open dynamic list editor |
||
|
List dynamic list in analysis window |
||
|
Open parameter editor |
||
|
Open browser to (un)load libraries; show
list of loaded libraries |
||
|
Open user GUI |
||
|
Open analysis launch window |
||
|
Stop and shut down analysis client,
disconnect analysis server |
||
|
Stop and shut down analysis server |
||
|
Start analysis. Monitor pad: start
updating all objects in list, or only
displayed ones. |
||
|
Stop analysis |
||
|
Open analysis configuration window (can
be closed/opened any time); browser popup menu: edit selected |
||
|
Open analysis output window (can be
closed/opened any time) |
||
|
Open file browser |
||
|
Open color editor |
||
|
Expand/shrink histogram in selected pad
in X. |
||
|
Expand/shrink histogram in selected pad
in Y |
||
|
Expand/shrink histogram in selected pad
in Z. |
|
|
|
Move expanded histogram in selected pad
in X direction |
||
|
Move expanded histogram in selected pad
in Y direction |
||
|
Move expanded histogram in selected in pad
Z direction |
||
|
Set fill color |
||
|
Set line color |
||
|
Set marker color |
||
|
Scale Y axis linear/logarithmic |
||
|
Scale Z axis linear/logarithmic |
||
|
Scale X axis linear/logarithmic |
||
|
Draw 1d histogram/line style |
||
|
Reset display in selected pad to
histogram limits |
||
|
Open window to set display limits
(applies to selected pad, or all pads if this option is enabled in view panel) |
||
|
Execute Tree draw. |
||
|
Kill analysis |
||
|
Clear button in browser pads clears
objects in analysis, in condition editor clears counters. |
||
|
Enable clear function for objects |
||
|
Disable clear function for objects ( does
not clear these objects) |
||
|
Analysis pad: copy selected object(s) to
monitor |
||
|
Remove selected object(s) |
||
|
Move selected object(s) to memory (from
analysis, monitor, or histogram server); or copy object from analysis to
editors (conditions or parameters) |
||
|
Copy object in editor to analysis
(conditions or parameters) |
||
|
Analysis pad: update folders from
analysis. Memory pad: update all objects from analysis and redraw. |
||
|
Browser icons for window condition
(arrays). Window mode in marker editor |
||
|
Browser icons for polygon condition
(arrays). Polygon mode in marker editor |
||
|
Browser icon for TCanvas |
||
|
Browser icon for TGraph |
||
|
Browser icon for Go4 pictures. |
||
|
Brower icon for TH3 histograms |
||
|
Browser icon for TH2 histograms |
||
|
Browser icon for TH1 histograms. Button:
draw selected objects (one per pad). |
||
|
Draw selected objects (all in one pad,
superimpose) |
||
|
Save selected object in memory to ROOT
file |
||
|
Refresh memory list (needed to see new
histograms created e.g. by ROOT in the GUI). In condition editor: refresh
values from view panel. |
||
|
Browser popup menu: open information
window for selected histogram or condition |
||
|
Editors: shows up if object in editor
differs from object in analysis (file). Use for update. |
||
|
Condition editor: connect to a picture
with conditions (gets list of conditions from it) |
||
|
Condition editor: update graphics from
values in editor. |
||
|
Output values of condition editor, info
window, or markers according settings in the log action. |
||
|
Close window without further action |
||
|
Browser icon for dynamic list entries |
||
|
Insert arrow in marker editor |
||
|
Pick next mouse click in pad to get
values into condition editor or marker editor |
||
|
Browser icon for a tree |
||
|
Browser icon for a branch |
||
|
Browser icon for leafs |
||
Note that the Alt-x keys work on windows whereas the Ctrl
(Strg)-x keys work directly. Sometimes the same
function is available in both, i.e. Alt-a-n or Strg-n.
In these cases the last character is identical.
Ctrl-O |
Alt-F-O |
File menu: Open local file |
Ctrl-R |
Alt-F-R |
File menu: Open
Remote file (TNetFile,
TWebFile, TRFIOFile) |
Ctrl-Y |
Alt-F-Y |
File menu: Save
all objects of memorY
browser to ROOT file |
Ctrl-Q |
Alt-F-Q |
File menu: Close
(Quit) all files |
Ctrl-X |
Alt-F-X |
File menu: EXit Go4 |
|
|
|
Ctrl-V |
Alt-T-V |
Tools menu: Open
new View panel |
Ctrl-F |
Alt-T-F |
Tools menu: Fitpanel |
- |
Alt-T-H |
Tools menu: Histogram properties window |
Ctrl-I |
Alt-T-I |
Tools menu: HIstogram creation tool |
- |
Alt-T-O |
Tools menu: COndition properties window |
Ctrl-C |
Alt-T-C |
Tools menu: General
Condition editor |
- |
Alt-T-E |
Tools menu: Event printout and inspection
tool |
Ctrl-D |
Alt-T-D |
Tools menu: General
Dynamic list editor |
Ctrl-B |
Alt-T-B |
Tools menu: Load
liBrary dialog |
Ctrl-U |
Alt-T-U |
Tools menu: User GUI |
|
|
|
Ctrl-N |
Alt-A-N |
Analysis menu:
LauNch analysis client
process |
Ctrl-M |
Alt-A-M |
Analysis menu:
Disconnect (reMove)
analysis client |
Ctrl-T |
Alt-A-T |
Analysis menu:
SubmiT settings and start
analysis run |
Ctrl-S |
Alt-A-S |
Analysis menu:
Start analysis run |
Ctrl-H |
Alt-A-H |
Analysis menu:
Stop (Halt) analysis run |
Ctrl-G |
Alt-A-G |
Analysis menu:
Show/hide analysis confiGuration
window |
Ctrl-W |
Alt-A-W |
Analysis menu:
Show/hide analysis output terminal Window
|
|
|
|
- |
Alt-S-O |
Settings menu:
shOw/hide... |
- |
Alt-S-F |
Settings menu:
Fonts... |
- |
Alt-S-Y |
Settings menu:
StYles... |
- |
Alt-S-L |
Settings menu:
Log actions |
- |
Alt-S-H |
Settings menu:
Generate Hotstart |
- |
Alt-S-T |
Settings menu:
Analysis Terminal history
length |
- |
Alt-S-S |
Settings menu:
Save Settings |
|
|
|
- |
Alt-W-S |
Windows menu:
CaScade |
- |
Alt-W-T |
Windows menu: Tile |
- |
Alt-W-C |
Windows menu: Close all windows |
- |
Alt-W-M |
Windows menu: Minimize all |
- |
Alt-W-O |
Windows menu:
Save LOg window to text
file |
- |
Alt-W-L |
Windows menu:
Clear Log window |
- |
Alt-W-A |
Windows menu:
Save Analysis window to
text file |
- |
Alt-W-W |
Windows menu:
Clear analysis Window |
|
|
|
- |
Alt-H-I |
Help menu: Read
Go4 Introduction manual |
- |
Alt-H-R |
Help menu: Read
Go4 framework Reference
manual |
- |
Alt-H-F |
Help menu: Read
Go4 Fitpackage manual |
|
Alt-H-G |
Help menu: Read Go4 GUI macro command reference |
F2 |
Alt-H-Q |
Help menu: About
Qt |
F3 |
Alt-H-O |
Help menu: About
ROOT |
F4 |
Alt-H-G |
Help menu: About
Go4 |
|
|
|
|
|
|
|
|
|
Alt-U |
- |
If analysis
configuration window is active: SUbmit
analysis settings |
|
|
|
|
|
|
- |
Alt-I-S |
View panel
file menu: Save as... |
- |
Alt-I-P |
View panel
file menu: Print... |
- |
Alt-I-O |
View panel
file menu: ClOse View panel |
|
|
|
- |
Alt-E-E |
View panel
edit menu: Show/hide marker Editor |
- |
Alt-E-R |
View panel
edit menu: Show/hide ROOT
attributes editor (TGedEditor) |
- |
Alt-E-C |
View panel edit
menu: Start Condition
editor and work on pad conditions (in pictures) |
|
Alt-E-E |
View panel
edit menu: Show/hide object Event status line |
- |
Alt-E-1 |
View panel
edit menu: Change to 1:1
coordinates ratio |
- |
Alt-E-D |
View panel
edit menu: Change to Default
pad margins |
- |
Alt-E-M |
View panel
edit menu: Clear Markers |
- |
Alt-E-P |
View panel
edit menu: Clear Pad |
- |
Alt-E-A |
View panel
edit menu: Clear CAnvas |
|
|
|
- |
|
|
- |
Alt-O-C |
View panel
options menu: Toggle Crosshair
mode |
- |
Alt-O-S |
View panel
options menu: Show/hide histogram Statistics
box |
- |
Alt-O-T |
View panel
options menu: Show/hide histogram Title
box |
- |
Alt-O-L |
View panel
options menu: Show/hide multiplot Legend |
- |
Alt-O-K |
View panel
options menu: Keep view
panel title |
- |
Alt-O-V |
View panel
options menu: Set View
panel title... |
- |
Alt-O-I |
View panel
options menu: Toggle SuperImpose
mode |
|
|
|
|
|
|
|
|
|
The following UML
scheme gives an overview of the event base classes and typical implementations:
The TGo4MbsEvent is filled from the TGo4MbsSource
(both provided by Go4). The TUserEventProcessor, which had been defined
to match the user’s experiment, takes the raw data from GSI format 10,1 and
unpacks them into the TUserEvent object. Both TGo4MbsEvent and TUserEvent
objects can be stored into (different) TGo4FileStore instances. Later
these can be read again event-by-event using the TGo4FileSource.
Analysis
class 22
client 24
framework 8, 21
launch 27
server 24
setup 28
step 22
Auto-save 14, 26, 49
restore 27
save 28
Browser 54
export 14
protection 12
remote 12
shortcuts 13
Condition 27
create 69
dynamic list 76
editor 16, 67
marker 13
marker editor 66
Dynamic list 76
condition 76
event 76
histogram 76
tree 76
Event
classes 21
MBS 21
print 13
Fitter 72
sigma 13
Folder 26, 54
user objects 13
Histogram
create 58
dynamic list 76
Hotstart 14
Libraries
.rootmap 12
load 45
path 17
rfio 12
userGUI 80
Macro
analysis 49
condition 27
GUI 81
parameter 26
path 9, 49, 81
Marker 64
condition 66
editor 13
Parameter 26
editor 26, 74
object 74
Picture
pad index 70
Rebin
monitoring 8
Tree
dynamic list 76
show 13
View panel 59
graphical editor 13
hotstart 14
legend 12
marker 13, 64
title 13