HOW CREATED
SctRodDaqRelease4.3.RC1 was created on 15th Dec 2005 from the tips of the following branches.
Package BranchTag VmeInterface HEAD RodDaq SCT_DSP_DEV_1 SctRodDaq TDAQ14_BRANCH
The state of each of the above branches at the time of the release was recorded with a single CVS tag: "SctRodDaq_4_3_RC1".
KNOWN PROBLEMS
- exampleSetup.sh and exampleSetup.csh
- Both of the above files incorrectly source a tdaq-01-04-00 instead of tdaq-01-04-01. This is fixed in CVS and in SctRodDaqRelease4.3.RC2 and in later releases.
- SctApi:
- Run enables settings not loaded from configuration database
- Dsp:
- In principle: "Stability in the face of errors during histogramming"
- In practice: "The DSP code was not checked into CVS and so the DSP code supplied with this release is a non functioning intermediate version. See SctRodDaqRelease4.3.RC2 for the corrected DSP code.
- Full scans - fail in analysis service and are not analysed AJB
- Analysis and book-keeping
- IGUI interaction
- Though CalibrationController now pays attention to RunType? (Cosmics/Calibration/Physics) in RunParams? (IS), there seems to be an IGUI bug that prevents it sometimes sending the correct RunType? to IS in the first place. Usually cured by =manually= selecting the correct RunType? in the IGUI RunParameters? tab. Chris
- DDC
- Though DDC was thought to have been demonstrated as working with this release on or around the 7th December 2005 (Bilge and Chris) it has more recently (14th December 2005) proved impossible to prevent the ddc_ct program dying about 3 seconds after starting -- not clear whether we are giving it a bad configuration, or whether ddc_ct has a bug of its own. Sometimes it seg faults, sometimes it gives an invalid operatation. gdb shows it dying when a dynamic_cast throws and exception inside one of the external PVSS libraries -- maybe we need to update the PVSS distribution to exactly the recommended version. Chris
- SctDataDisplayer?
- Some kind of compiler bug prevents compilation of DataDisplayerMain?.cxx in "-O3" mode, though lower levels of optimisation are OK. By default, releases are built in "-O3" however I have relaxed this to "-O2" for DataDisplayerMain?.cxx to work around this problem. Ought not to matter to end users. Chris
- FlashLoadCrate?
- Probably only in the relase as a binary! Sourcode should be in cvs somewhere though. JCH?
- ( Fixed in SctRodDaqRelease4.3.RC2 -- Chris )
- NoiseOccupancy? and FullBypass? scans sometimes slow
- On both occasions I observed this, the partition name was something other than "SCT". Was this just a coincidence, or is there a problem with the partition name lurking in there? Chris
- sctGui Dave
- The TIM Register gui (from right-mouse click on the TIM on the rod crate display) is not yet ready... so a dialog appears to that effect.
- Depressing mail from Peter
Subject: More gremlins, all pretty serious :-( Hi, This is to let you all know of a couple of problems we discovered today. 1) One of the modules appears to be sending no repsonse today, hence we have only 11 links. 2) with SctRodDaq_TDAQ14, as installed at SR1, the analysis server dies as soon as a scan finishes. Not good. 3) with SctRodDaq_0104, _0104_P and _TDAQ14 using our latest version of the DSP code, nmask scans work with the remaining modules in a single group. If we run 2 or 3 groups, the scans die due to BC errors at the first event. This is very strange, we cannot think of any changes to the DSP code that might have caused this. Most of this work was performed using the latest firmware loaded to slot 10: we were able to backtrack the router and efb but not the ROD controller. However the ROD in slot 11 has older firmware, and still gives the same result for three grups (last test). I don't suppse there may have been a little tinkering of BC offsets in all three versions? Well, looking at the register contents, apparently not. We can confirm that if we set the ROD register bit to suppress BC errors, scans with multiple groups work normally. After one of the failure I recorded a verbose probe and was surprised to find that all 6 modules had the same BC count irrespective of their group. So it isn't something to do with resets being sent to each group at different points in time. What changes as the number of groups is increased? Any ideas? I gues it's been a while since we tried to run with multiple groups.... Cheers, P.
- SystemTests
- config/databases/SCT_test.data.xml has not yet been ported to tdaq-01-04-01 and is not yet part of this release. As a consequence, the system tests will probably not work unless you get hold of a copy from an older version. Chris
KNOWN GOOD FEATURES
- TDAQ compatibility!
- SctApi:
- Automatic switch to physics mode when run type selected
- Dsp:
- Cause of errors in events now reported to text buffers
- Scans
- Fixed Frequency Veto
- RodDaq?
- As I understand it, the SCT_DSP_DEV_1 branch works. I'll see if there's anything obvious that's missing from it. Bruce
- SystemTests/CDITest
- ISSummaryWriter?
- Initial comittal of the ISSummaryWriter?. It writes the NoiseOccupancyTest and the NPtGainTest? summaries natively into IS. Defects soon to be added. Bilge
- Pipeline now has a STUCK and DEAD cell mask and now tells you which cells are stuck is now in the PipelineSummary?. Bilge
- sctGUI Dave
- I think sctGUI (as checked into TDAQ14_BRANCH) is ok for the release.
- many bug fixes, eg:
- IPC refs now seem ok when cycling daq thru boot state
- JCheckBoxes? removed from menus (these stopped working with java 1.5)
- new ROD crate display seems to work fine
- No known bugs/features (probably because its not tested exhaustedly! :-) )
OTHER COMMENTS
- eventStreamDecoder
- OK. Bruce
- OKS configuration
- I haven't looked at recently... It should work, but I've changed the schema a bit since the last revision of Shaun's converter. I'm happy for it to go in the release though. Bruce
See also SctRodDaqRelease4.3.