DAL sessions in Victoria

[Rob Seaman]

> Every dataset is labeled by date-string, eg 20060425 means today.  
> And so the regular expression "200604*" means everything from April  
> 2006, ".*" means everything, or you can be specific with "20040423| 
> 20040522" means anything from either of those dates.

This requires some care.  One is typically interested in retrieving -  
say - all images from a particular proposal, instrument and telescope  
on a particular local night as defined by the observing schedule.  On  
the other hand, timestamps such as DATE-OBS are usually expressed  
using universal time, so a timezone (and conceivably a daylight  
saving) correction is required.  In addition, one has to accommodate  
observatory operating procedures in designating a midday pivot  
dividing one night from the next.  NOAO uses noon, but there are  
undoubtedly instances in which daytime calibrations begin in the  
morning or end in the afternoon.  All bets are off if we aren't  
discussing ground-based nighttime instrumentation.

> Another alternative is to give an exact time (eg MJD) to each  
> image, and specify a time interval that way.

Suggest that humans find ISO 8601 easier to understand than MJD for  
the purpose described above.  That certainly has been our experience  
since 1993.  We currently cache our holdings into directories labeled  
by observing calendar date, for example, tonight's data will be  
written into .../20060425 and will have DATE-OBS values extending from:

	2006-04-25T19:00:00 to 2006-04-26T18:59:59 (inclusive, but with some  
clock jitter)

The 19 hour offset is 12 noon plus the seven hours from Arizona to  
Greenwich.

We use the UT timestamp (a heuristic tree starting with DATE-OBS,  
ending with the Unix NTP clock) to also generate a (near) unique  
OBSID keyword to label our raw data holdings, e.g.:

	OBSID   = 'kp4m.20060425T114910' / Observation ID

This is prefixed with a telescope identifier, but is sometimes  
disambiguated with instrument name or a trailing running number for  
instruments that may take multiple exposures more rapidly than every  
second.

Rob
seaman at noao.edu

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.ivoa.net/pipermail/dal/attachments/20060425/5f8d19cf/attachment-0001.html>