ScienceTools: Not covered at the meeting, but you can read the most recent report on your own.
FSSC Report: (John, Eric) are chasing the last few problems to be addressed for the new release.
Overlay (Leon) found one last bug and fixed it. (Tom) has relaunched the Overlay job; it's nearly done.
Pass 7 (Richard) Luca has created a To-do list. Leon is working on several items, such as adding sign information to rocking angle, a Jonathan Ormes request. (Anders) The livetime information to be added will just be the two 64-bit counters produced by Flight Software. They're already in the SVAC tuple.
(Toby) Speaking of merit variables, we need to clean up the use of multiple copies of pointing information algorithms. [See e.g. discussion back in May of last year under Interleave subtopic. ed.] Currently there is one in FluxSvc and one in merit. The latter is preferred since it can be used for both simulated and real data. (Heather) Who will take care of this? (Toby) will modify standard jo files such as basicOptions; other special-purpose files will have to be modified by whoever uses and maintains them. He's not proposing modifying any code now. Could be done by just taking out the line which lets Gaudi know about the algorithm in FluxSvc. (Joanne) would like to see a schedule for the code excision as well. We won't really know if we've caught all uses of it in job options until this is done.
(Tom) is putting together the Pipeline reprocessing task. Basic outline is in place, but there are many blanks to be filled in. Eventually will also want to reprocess FT1, but merit will be first.
Disk space (Richard) We're running out again, down to 22 Tbytes. Tony will be freeing up an additional 9 Tbytes soon, but that won't be enough to last long. More disk is on the way, but we don't have an ETA, nor do we know how long it will take the Computer Center to install it once it does arrive.
GR tags (Heather) will be merging all Overlay tags into the main branch.
v16r0 has the new ROOT version. It's passed System Tests; still need builds for Mac [see details from Emmanuel below] and RHEL5.
Documentation: (Chuck) sends this summary:
I've reinstated Gleam Outputs page in order to provide ready access to the documentation describing the Standard AnalysisNtuple Variables and CTB Variables, as well as to the CAL and the SVAC Tuple variables. It is now linked in from the green data access navbar, as well as from the Gleam diagram. I've also posted the pictures from Ken Fouts 'going away' party to the ISOC's public 'Photo Gallery' page. The pictures from the AAS meeting have also been posted and linked in from the SAS/Collaboration web page's 'Picture Gallery'. In addition, both of these photo albums are linked in from January's Newsletter, which is now on the streets.I also need to make a new tag for the workbook, so Berrie can update the mirror site, and I'm currently continuing to research the legacy TKR, RECON, and CAL software, which will be an ongoing project for the next few weeks.
cfitsio/tip (Toby) The tip test program for the latest tag (not yet in any ScienceTools release) is failing on Windows only. He first noticed a problem while working on other code using tip. The failure has to do with support of large — index requiring more than 32 bits — arrays in cfitsio. Under some circumstances the top 32 bits get trashed. It appears to be Windows compiler bug concerning templated virtual functions specific to Studio 2003; it's ok when compiled with Studio 2005. James Peachey suggests using cfitsio v1r3060p1, but this version is already being used in ScienceTools LATEST builds.
(Heather) had gotten an ok for a new ScienceTools release including the new tip, new cfitsio and incidentally a new ape, but now that needs further examination
(Tom) wonders if the newer version of ape addresses the problem he has with limits on lengths of inputs. In particular, file paths he needs to supply to gtexposure can exceed this limit. gtexposure itself has an algorithm bug which may be fixed in the proposed new ST release.
u30 (Heather) u30 has been cleaned out, thanks to Navid's efforts. It had filled up, causing RM problems, while he was on vacation. We need a better strategy for foreseeing and avoids such problems.
ROOT Mac build (Emmanuel) reports
Initially the ROOTv5.20.00-gl1 build was failing due to ROOT's configure script deciding to use Fink. Fink is only available for Mac power pc architectures, but we need ROOT built for an Intel Mac. There is no explicit "configure" flag to disable the use of Fink, but I was able to hide Fink (by taking the directory containing it out of the list in in $PATH) so that the configure script cannot find it. This appears to have worked: libraries and binaries have been built. However, I still get the following output:Error: string() declared but not defined prec_stl/vector:422: Error: string() declared but not defined prec_stl/list:187: ./configure: line 1: freetype-config: command not foundThe last message about configure not finding freetype-config could be avoided, in principle, with "--disable-builtin-freetype."
I am looking into the "string() declared but not defined" errors now. Although the root libraries and binaries have been built despite those errors, I have yet to install ROOTv5.20.00-gl1 in its official location, ...GLAST_EXT/tiger-i386-32bit/root. I want to first resolve the above errors.
ACD mini-review (Heather) Hot-off-the-presses note from Richard indicates it will be Wednesday (tomorrow) at 10 Pacific.
SCons, RM (Navid) Optimized builds are enabled. Windows builds are not going yet. Mac builds are down because of lsf miscommunication (it thinks jobs are running when they actually aren't). The LSF daemon was locally-compiled (by Computer Center) and apparently isn't quite right.
In other news, he separated out the little bit of container-specific code in SConstruct into a new per-container file. The rest, still called SConstruct, now lives in the SConsFiles module in CVS, shared by all containers.
SCons and Windows (Joanne) The saga continues. Main points since last time:
CHS (Joanne) No change from last time: ready to go pending ok from Navid and Bryson.
Python 2.6 (Toby) is not eager to upgrade just yet since numpy — make that pylab — is not yet available for 2.6. (Tom) would still find it useful to have 2.6 installed somewhere for testing.
previous | minutes index | next |