Details
-
Type:
Story
-
Status: Invalid
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: obs_decam
-
Labels:
-
Story Points:4
-
Epic Link:
-
Sprint:Alert Production F16 - 11, Alert Production F16 - 11b, Alert Production F16 - 11c, Alert Production S17 - 12, Alert Production S17 - 1, Alert Production S17 - 2, Alert Production F17 - 9, Alert Production F17 - 10, Alert Production F17 - 11
-
Team:Alert Production
Description
Processing Decam data from raw images currently requires downloading the bad pixel mask files from NOAO, then ingesting them into the calibration repository. It would be much more convenient if we included these with obs_decam itself, pending the ok to do so from NOAO. Our understanding is that we could create a default calibration repository in obs_decam, which would be the parent of users' calibration repos. We will also likely want to convert the provided files from mask images into tables of bounding boxes.
Attachments
Issue Links
- relates to
-
DM-14866 decam defect ingest silently retrieves calibDate from file path
- Won't Fix
-
- Done
-
- Done
-
- Done
-
- Won't Fix
Activity
Resurrecting this thread to remind, uh, myself that we ultimately want to be able to ingest decam defects with mode=link (or move or copy). The pipe_tasks ingestCalibs default is presently move, but it's necessary for now to ingest defects explicitly with mode=skip since they don't have the OBSTYPE header keyword that pipe_tasks' getDestination expects.
Bump. This issue is still really annoying and causes all sorts of bewildering errors if one doesn't ingest the defects just right.
For instance, since the default mode for ingestCalibs.py is now move, defects can't be ingested unless one knows to add "mode=skip" (even though the error message is complaining about bad headers; Meredith Rawls's comment was the only way I figured this out). Next, the mapper now throws an error (during processCcd, not even on ingest) if the templated path is absolute rather than relative to the repo root. This means that one has to either copy the defects into the repo before ingestion, or ingest with a convoluted relative path to get to a fixed location outside the repo. This is all way more pain than it should be.
Colin Slater
added a comment - Bump. This issue is still really annoying and causes all sorts of bewildering errors if one doesn't ingest the defects just right.
For instance, since the default mode for ingestCalibs.py is now move, defects can't be ingested unless one knows to add "mode=skip" (even though the error message is complaining about bad headers; Meredith Rawls 's comment was the only way I figured this out). Next, the mapper now throws an error (during processCcd, not even on ingest) if the templated path is absolute rather than relative to the repo root. This means that one has to either copy the defects into the repo before ingestion, or ingest with a convoluted relative path to get to a fixed location outside the repo. This is all way more pain than it should be. Just read Meredith Rawls's last comment – the underlying bug with OBSTYPE is DM-13975.
Krzysztof Findeisen
added a comment - - edited Just read Meredith Rawls 's last comment – the underlying bug with OBSTYPE is DM-13975 . This is finally being addressed by DM-19730! Rejoice!!
A somewhat relevant thread: https://community.lsst.org/t/storing-an-instrument-model/679
Maybe it's just me being confused; I wonder if this is something for RFC.