Teraminutes for the meeting of Friday 7th January 2005

Present: Yannick, Gilles, Henry, Mireille, Jean-Christophe, Emmanuel, Fred

Absent: Laurent, Gerard

Skywatcher (Jean-Christophe):

 VOTABLE functionality implemented. Jean-Christophe developed a C-library VOTable. This is the only C-lib done so far (others are in C++, PERL, JAVA, etc..) and it should be made public to AVO users. Yannick asks Jean-Christophe to prepare a package that will be delivered as a Terapix AVO contribution.

 Yannick asks for further details so they can be included in the AVO document. This is a more general request to all: please send all AVO developments done to Yannick in order to prepare a general summary.

 Jean-Christophe reported difficulties in using HTM that seems to take unexpectedly long time to run with MySQL. Jean-Christophe also reported problems during compilation with the HTM Lib.due to declaration conflict with MySQL. These problems should be fixed soon with the help of Anthony.

QualityFITS (Fred):

 Many images were found to be completely blank; we are not sure why these were accepted by the CFHT These images caused QualityFITS to crash and all SPICA processing to stop.

 Fred has since patched QualityFITS so that it can continue without crashing on these images.

 PSFex produces a value of 0.65 even with blank images (don't know why this happens ). Emmanuel will investigate

 Packaging and documentation ready for the end of next week.

Defectix (Anthony):

 Yannick tried to make some tests following a simple rule: run it, stop when it crashes, send bug report to Anthony, wait for bug fixed, and run again, etc... Yannick asks Anthony to create a default configuration file that includes all information we need to run Defectix on MEF and Swarped Megacam images just by using the "defectix -M computing" command.

 Anthony will provide this to Yannick so that tests can continue.

ICD (Interface Control Document) (Yannick, Fred, JC):

 Yannick is waiting for input from Emmanuel and Laurent. First Emmanuel, then Laurent.

Science cluster (Fred):

 Fred has installed the operating system on cosmix2 and also many other scientific packages.

 Machine seems to work well. No stability issues. The machine is around 2x faster in fp than cosmix.

Stability of production cluster machines (Fred, Emmanuel)

 The pipeline has processed over 8000 images with only one hardware crash. No crashes on mix2, but it is not clear that if is due the new Pann-starrs kernel config or not (mix1 has not crashed either, and the kernel has not changed there).

Spica (Laurent):

 Laurent was not present for the meeting. Some discussion about management of the pipeline, and the need to provide a better means to follow the processing of the pipeline. There will be another meeting at the start of the week where these matters are discussed in more detail.

 Yannick pointed out that the top priority for SPICA and T0002 is the wide survey. The experience acquired by processing T0001 showed that we do not envision any critical issues with the Deep. In contrast the wide may still have unexpected problems during SPICA processing.

Backups (Gilles):

 Gilles is currently backing up the disks on all the alphas. The idea is that the alphas will be turned off in the third week of January. One remaining problem is that the alphas are being used because there is an IDL license installed there.

Other matters (All):

 We all agreed we failed in using "dotproject", except Laurent Domisse. Please do use this tool for all tasks, even if you use emails in addition.

 Yannick ask why the http://clix.iap.fr/steeringroup/seeing_ccd/ page is empty and does not contain stats on seeing anymore. Jean-Christophe will look at this. Apparently the other stat information are still there.

 Emmanuel is currently writting the documentation for STIFF. Some improvements have been made by Emmanuel to better handle saturated data on 3-color images.

 mix4 will not be installed in the cluster for this release, but will be integrated for the next release, at which point mix2 will be removed from the cluster. The opteron bi-pros will be used for processing outside the pipeline.

 Photometric fields for the wide. To make the global photometric solution over all the survey fields, we need many images. However, with the default threshold there are too many objects in the catalogues and consequently not enough memory to manage these catalogues. In this case, to compute the photometric solution we should raise the S/N threshold in Scamp. This is being implemented in SPICA.

Subversion (Anthony):

 Subversion is a new CVS-like too. Links in the spip pages installed. Under testing.

SCamp (Emmanuel):

 Mireille reports that the orientation of the distortion diagram is reversed with respect to the input fits images. Emmanuel will fix this adopting a single N-E orientations to all images.

Grading images (Mireille, Henry, Yannick):

 Still many images to grade. If someone grades images, they should be careful only to grade images on a certain machine, in order to avoid conflicts. Mireille will grade images starting from July 23, 2004 on mix3. Henry will grade images starting from July 23, on pix9. Yannick will grade images during the week end.

Next tera-meeting: Thursday 13th January 2005