Choose timezone
Your profile timezone:
ROOT Team Meeting
→
Europe/Zurich
32-1A24
32-1A24
-
-
11:00
→
11:01
News 1mLorenzo request for a fellow see report by Christian Holm about Warnings with gcc4.2 (and solaris too). MESSAGE FOR FONS ============== Install SVN 1.4.4 (problems with 1.1.4 on SLC4 today) MESSAGE FOR SPI ============= Install gcc4.2 I will be away between November 8 and 17 (school in Hanoi). I gave a seminar at DESY on Monday. "Software Challenges for next Decade" My slides are available. at DESY I met Sergey Chekanov. he has developed a "better JAS" in Java. see: http://projects.hepforge.org/jhepwork/page4.php Main message: he would like to cooperate with us. He is now moving to BNL to work with Atlas. NOW: Main topic of this meeting is QA ============================= Each work package manager takes the responsability for QA of his area. Ideally he/she should make a web page with a checklist of items validated and on which machine. Release 5.14/00 quite poor; A lot of patches Release 5.16: Extremely bad. In fact unusable and dangerous (make install severe problem) Release 5.17/04: Severe bug with TBrowser release 5.18 likely to be the production version for 3 LHC experiments for 1 or 2 years. WE CANNOT MISS THE BOAT
- 11:01 → 11:03
- 11:05 → 11:07
-
11:10
→
11:15
QA plans: Roofit 5mSpeaker: Wouter Verkerke> Hi Wouter, > > > > This morning we have our team meeting discussing QA with the coming release. > > Could you send me briefly your plans > > -stressRoofit.cxx Hi Rene, I plan to implement O(20) tests that test fitting, plotting, persistence, event generation and all integrate them in the stressRoofit driver. I have O(15) tests as loose macros right now that I use for private release testing, I need to add another ~5 or so mainly to test persistence as that is a new feature. I intend to produce the standard output format that other core tests currently do > > -validation with valgrind Will make this part of the routine before new releases. I think it will be the easiest to run root with valgrind on the above set of tests as that covers most functionality. The time scale for both is ~2 week (next week is lost because of atlas t/p week) Wouter
-
11:15
→
11:20
QA plans: CORE 5mSpeaker: Fons RademakersCore ==== -systematic tests of make/make install with different --prefix configs. -test the make install procedure starting from the tar file + svn -test when starting from scratch or after a svn up. -if possible tests with different compiler versions (in particular gcc4.2 seems to be a big requirement) -tests the results of make dist by checking if the corresponding tar (or .msi) files can really be used on the target platform. -test the size of the tar files after make dist or make distsrc to make sure that unwanted files are not written to the tar file. -test the make rpm or debian with validation
- 11:20 → 11:25
- 11:25 → 11:30
-
11:30
→
11:35
QA plans: Graphics 5mSpeaker: OlivierGraphics ======== -Extension of the existing stressGraphics with at least 40 more tests including all existing graphics objects -test batch mode generation of ps, gif, jpg, root, etc -test with different color palettes -test of DrawClone and DrawClonePad -test of SavePrimitive -include more tests with asimage -all tests passing valgrind -test for memory leaks -all tests running on all supported platforms MODIFY stressGraphics such that hsimple.root is generated automatically if not yet there From Olivier ========= The extension of stressGraphics started. There is 37 tests right now. I am setting up the memory leaks tracking. My idea is to do it the following way: MemInfo_t gRAM0; MemInfo_t gRAM1; test1(); gSystem->GetMemInfo(&gRAM0); for (int i=1; i<=10000; i++) test1(); gSystem->GetMemInfo(&gRAM1); Int_t leak = gRAM1.fMemUsed - gRAM0.fMemUsed; Doing that on the very first test in stressGraphics (simple TLine example) I get gRAM0 = 427 and gRAM1 = 522 so leak = 95MB. If I do the loop 1000000 times "leak" increases. Oncee the leak test will be completed I will start on the other QA items: - More tests in stressGraphics (with PS as reference). In particular different color palettes, DrawClone and DrawClonePad, more on asimage - Generation of gif, jpg, root, macro etc files ... - Execute back the generated macro files (SavePrimitive) to check if they produce the right result. Then, once the stressGraphics will be more complete it is should be run on all supported platforms and also with valgrind.
-
11:35
→
11:40
QA plans: GUI 5mSpeaker: IlkaGui --- -A procedure/checklist to test systematically all widgets, eg how to detect the latest big bug in TBrowser when clicking twice on the same object. -A procedure to test the new browser -test of the builder and code generator(CRTL/S) -systematic tests of all object editors -compiled and interpreted tests -all tests passing valgrind -test for memory leaks -all tests running on all supported platforms
- 11:40 → 11:45
-
11:45
→
11:50
QA plans: MATH 5mSpeaker: LorenzoMath: ===== development of $ROOTSYS/test/stressMath with about 50 tests along the lines of what we already do in stressLinear, stress, etc. This should test -I/O with smatrix and LorentzVectors (with gccxml and rootcint) -90% of tests executed with CINT -all tests executed with ACLIC -all tests passing valgrind -test for memory leaks -all tests running on all supported platforms
-
11:50
→
11:55
AOB 5m
-
11:00
→
11:01