Minutes of the terameeting of Friday August 5th 2005

Present: Mireille, Chiara, Jean-Christophe, Anthony, Laurent, Emmanuel, Yannick

Status of T0002 data (Mireille, Yannick, Laurent)

 Mireille found the problems raised by Hervé Aussel regarding the release candidate. At least one image has been graded "C" and should be removed from the release and the release candidate sent to CADC.

 Laurent looked at the problems regarding the Pre-W data also raised by Hervé. Two different problems:

• some data with RA "08 h" have been obtained before June 1st 2003. About 20 images have to be removed from the CADC T0002 release.
• some data have wrong DEC (+00:59 instead of -00:59). Laurent contacted CFHT about this issue since it seems it was a CFHT mistake.

 All wrong images must be removed from CADC data base as well as the release and the release candidate archived at Terapix. Some new images must also be archived in the release. Jean-Christophe reminds that new images must be first checksum-ed and is not sure CADC can remove images easily from their data base. Jean-Christophe, Fred and Yannick will meet on Monday August 8 in order to summarise all immediate actions needed.

 Jean-Christophe reported there are still some T0002 files missing at CADC. Some .ps files have apparently wrong names but first checks done by Jean-Christophe do not confirm there is any mistake.

Status software (Emmanuel)

 SExtrator: Not much progress. Emmanuel is still working on bugs reported by Mireille on the most recent SExtractor release. Fixing these bugs is the top priority since it will be implemented in QFITS-in for T0003.

 With the new SExtractor, parameters derived for each object are different in each filter, so catalogues derived from the chi2 image is not sufficient. It is likely we must change finale catalogues in order to provide a much more parameters than before (a set for each filter). Decision must be taken before August 25 on this issue.

 Need to check Fred uses WIN parameters in the new Qfits-in/out releases.

 SWARP: Emmanuel still found swarped images with inverted lines. This is only produced when using the multi-thread option, during co-addition.

 SCAMP: Emmanuel' attempts to get theoretical internal rms astrometric errors are still in progress. He succeeded on simulated images but failed on Megacam data. Latest tests show huge residual errors that suggest a bug remain on SCAMP.

 Emmanuel, Laurent and Jean-Christophe will check the internal UNSOB-1 option works.

T0003 data set: Up to now, 16264 megacam images have been transfered to Terapix. The most recent CFHTLS data have been obtained by May 15, so we expect to transfer all 05AQ05 in order to get all CFHTLS images up to June 15. If so, we will follow SAC recommendations (all data older than 6 months should be released by Terapix), assuming T0003 is released by Dec. 15, 2005.

With more than 16000 images , the processing time will be 20% longer than expected. In order to avoid delays, we will not grade PI data after QFITS-in, but only CFHTLS-related (L01-L06, Q98, Q99). SPICA will start at this stage. PI data will be graded later, when T0003 will be running.

Status SPICA: Laurent has made all top priority SPICA modifications requested for T0003. He will start extensive tests after August 15 (back from vacations).

 Laurent is now implementing new options (modifications of stack priority levels during the release construction; possibility to run stacks on free nodes).

 Laurent was wondering why we should run FITSFERIFY at the end of each stacks since we may be stuck anyway in case of error. This is true, but this is also true when FITSFERIFY is run at CADC. In all cases we must understand why error is detected and then run SPICA again. Running FITVERIFY after SPICA at TERAPIX will save time and will (hopefully) produce more reliable data sets at the end of the TERAPIX pipeline.

 Laurent ask to remove dbterapix T0002 data base. Since no one use it, we agree Laurent can do this immediately.

 Laurent failed to transfer QFITS-in located on mix nodes to ftpix. Need help from Fred.

Data transfer Jean-Christophe is still testing data transfer.

 Jean-Christophe reminds us there are still 1TB of temporary files on in/out FC bays that Fred should send to CCPN in Lyon. Since disk space is now missing (16000 images transferred, more coming) it is important these data be removed from the disks asap. Need to talk to Fred about this next week.

Web Service: Jean-Christophe did make good progress. A cgi application for graphic interface is working now.

EFIGI: Not much progress done because Anthony was out for most of the week. Emmanuel had feed back from Pascal Fouque who is ready to start classification. Pascal would like to have a printable output summary of his classification in order to ease cross check with RC3. Anthony will provide the output file.

Other matters Yannick will send Chiara several useful informations about WIRCAM, including the pwd for restricted access WIRCAM pages where tests images should be located.

 Please register to ADASS if you plan to go.

 The 4 new 2405 FWP arrived today.

Next meeting Friday August 12, focused on Fred activities.