We certainly can easily add more of the gen2 card block as seen in the PFSA files, which would include most of those cards.

But I think a basic question is whether we just need to add a few more cards or whether the required card list is actually the full Common list in the NAOJ Subaru Telescope Basic FITS Header Keyword Dictionary. If that Common list only applied to files with image data we would be fine, but it currently is required from all FITS files, and I think is what the STARS acceptance scripts are based on.

That declared WCS-ORIG requirement worries me – does that imply that we will be asked for a WCS? Since we have no image data even a false WCS makes little sense – what would CRPIX1 refer to? Is that even legal?

None of the NAOJ-approved values for DATA-TYP look right to me. Since we are archiving one PFSF per PFSA file we could use the SPS DATA-TYP, I suppose. "None" has been suggested, but is not in the NAOJ list.

For SPS-aligned PFSA visits we could add the expected EXPTIME. We could also add HST/UT/LST timestamps which would be a handful of seconds off the real value, or could describe when the cobras were moved. But these cards are getting far from the real pfsConfig content....

I know that we want to quickly help this process along, but can we find out what the final requirements are before we do any further work? The requested list is, after all, a list we got after implementing INSTRM-1863

From follow-up conversations with Pyo-san and the Subaru FITS committee, the following 7 cards are requested as minimum requirement to archive. :  RA, DEC, DATE-OBS, EQUINOX, DATA-TYP, PROP-ID, FRAMEID

• RA, DEC: the command (design) positions are set although Subaru's definition is the real positions. FITS committee allow us to keep command value (=designed value) but told us to put a comment that these are  command value . Unit should be sexagesimal.
• EQUINOX: as long as RA,DEC is used for pointing telescope, they ask us to set "J2000" (fixed).
• DATA-TYP: set META_{TATA-TYP of PFSA file}
• (DATE-OBS, PROP-ID, FRAMEID: already implemented)

DATA-TYP and EQUINOX added to ics_utils.

Any changes to RA/DEC would be done to datamodel, not ics_utils. I will point out that any conversion to sexagesimal would be hidden behind the datamodel persistence code – it can be done.

ics_utils tagged 1.6.11

If we will need RA and DEC formatting changed to sexagesimal, that should be filed as a datamodel ticket.

Just for recording, STARS will accept RA,DEC in deg format. But we still need to modify the comments.