Fix Version/s: None
Component/s: meas_base, pipe_tasks
Team:Data Release Production
See the last few comments on
DM-29528. Something is going wrong in either MergeMeasurements or ForcedPhotCoadd that's resulting in some kind of source ID inconsistency.
Jim Bosch added a comment - - edited
Jim Bosch added a comment - - edited This problem has turned out to be more than just DM-29737 could fix, but further work and discussion has moved to DM-29907 , so I'm closing this as a duplicate.
DM-29737should fix this, but I'm going to keep this open until I have a chance to understand why the catalogs aren't sorted (see https://lsstc.slack.com/archives/C2JPMCF5X/p1618428539388800), as fixing that is worthwhile too - it will address the (probably tiny) performance regression the DM-29737fix along would imply, and if we don't see this problem in Gen2 (as seems to be the case as per https://lsstc.slack.com/archives/C2JPMCF5X/p1618448136422900?thread_ts=1618428676.389500&cid=C2JPMCF5X), I bet tracking it down will help with Gen2<->Gen3 parity.