Ref Common Pedigree

[tddesjardins]

The ref_common schema pedigree enum only contains GROUND, MODEL, DUMMY, SIMULATION. We will need to add INFLIGHT and provide a capability to specify the date range. I'm not sure how JWST has done this — if I look at one of their ASDF reference files (abvegamag) on CRDS, I see pedigree is "INFLIGHT 2022-07-08 2023-03-18". This is only for reference and not used to my knowledge by anything downstream, so if we want to keep the enum but add INFLIGHT, we could break the date range information to an additional keyword.

This is obviously not urgent as we are not near launch, but I wanted to capture this before I forgot about it.

[schlafly]

The date range would be ~redundant with the CRDS use after information, but included so that the files are self-contained? Or it contains a beginning and an end date, while the use after information only includes the former?

[stscijgbot-rstdms]

This issue is tracked on JIRA as RAD-179.

[tddesjardins]

It's best to think of them as decoupled from USEAFTER. In practice, they are usually related, but they do not have to be. The PEDIGREE date range is supposed to indicate the range of dates over which the input files were taken to make the file. USEAFTER could be anything in theory, though is usually a datetime corresponding to the beginning of the range. For example, I could have darks taken on 10/8 - 10/13 but set the USEAFTER to 2024-10-07 00:00:00.

I'm not sure what JWST did about this...maybe @stscieisenhamer could advise? We could separate the date info from PEDIGREE if we want to keep the enum list. If we want the dates as part of pedigree, I'm not sure how we can keep the enum.

[schlafly]

Okay, that makes sense, thanks. It does feel more natural to me to give dates or MJDs as additional metadata than to shove them in the string, but I think we should treat these files as yours and so if you'd rather put them in the pedigree then we'll just make it a free form string.

[tddesjardins]

Asking ReDCaT about it, they are pretty adamant that the dates should be in the pedigree. In that case I think we need to take out the enum and leave this as a string for validation.