Teraminutes for the meeting of Friday October 29th 2004

Present: Everyone, with the exception of Mireille and Laurent.

Status of the release:

 Yannick summarises the current status of the release which HAS to be ready for the 5th of November, the date of the WIRCAM meeting. A second constraint is the upcoming SAC meeting shortly afterwards in Hawaii. Canadian astronomers are not happy and complain they have no CFHTLS data yet.

 We will release the four deep fields (d1,d2,d3 and d4). We will make TWO releases, one with astrometrix and one with SCAMP/swarp. The scamp/swarp/spica version would be preferred because all the metadata information is included and because is uses scamp which is much faster and usre friendly than astrometrix.

 However, continuing problems with spica seem to indicate that it's not certain that spica will be ready for the release date.

 Currently scamp works in spica, because the selection of images has been changed: many spurious images have been rejected, and fourth image class 'D' has been added, indicating images which cannot be used with scamp. The current selection for the deep release is:

 Data obtained AFTER May 31, 2003 and BEFORE July 27, 2004.
 Data class A or B
 Exposure time > 60 sec.
 Seeing < 1.101 (except for u: < 1.401)
 Airmass < 1.4

Stability problems on the cluster:
 Yannick summarises for the current state of affairs: mix1 and mix2 are still unstable, and there some problems with speed on mix3 for large swarp jobs (the slow-down always happens during the coaddition phase). Fred will try upgrade the glibc version on mix3 to resolve this problem. Still no idea why mix1 and mix2 still crash in a random fashion.

Spica problems: We are not sure why spica produces the correct photometric scaling. Although two images with the correct photometry were generated (which Henry verfified by comparing with the manually generated stacks). Also the meta data (plots and quality-control graphs) is not produced. Laurent will try sort things but his currently indisposed: JC and Fred will try understand what's going on, Laurent will try to restart working on this asap, likely this Friday afternoon.

64-bit qualityfits on mix2: Fred has produced a new version of qualityfits which works on the mix machines. Some problems with visual appearance of the plots, which are being investigated.

Skywatcher (JC): Work currently in progress on the parser library.

dbterapix (Gerard): Work in progress on the panier. Gerard is no longer paid officially after Oct. 31. He will then work 3 days a week for terapix , beeing paid by travel expenses.

Hardware: mix4, the second four-processor opteron system, should be delivered in the next few weeks.