Pinned fields
Click on the next to a field label to start pinning.
Details
Details
Assignee
Yusra AlSayyad
Yusra AlSayyadReporter
Yusra AlSayyad
Yusra AlSayyadStory Points
2
RubinTeam
Data Release Production
Checklist
Checklist
Created June 17, 2022 at 5:34 PM
Updated January 31, 2025 at 11:38 PM
Last summer when putting together the DRP DIA pipeline, the simplest way to make a goodSeeingCoadd was to write out a list of the best seeing visits (` BestSeeingSelectVisitsTask`) as a StructuredDataDict called goodSeeingVisits and then use goodSeeingVisits as input to assemble. Assemble’s runQuantum filters the input warps provided by the butler before passing them on to run. The QG is blissfully unaware of this down-select. This was fine for DP0.2 because we wanted to make deepCoadds too and thus warp everything anyway.
In experiment-mode, folks are going to want to make specialized coadds, won’t ALWAYS want to warp all the visits before coaddition. There is still no way for the QG to be aware that we only want goodVisits: https://lsstc.slack.com/archives/C01FBUGM2CV/p1655224983994479
So let's just give makeWarp the same treatment as assemble. Behavior might be a little confusing, since you will either get an empty warp or a "no work found" dep on
doWriteEmptyWarp
but as long as its consistent, it'll be better than nothing.