@sbailey I made this draft PR because I'm going to be unavailable for part of tomorrow and may not have time to change the code so this is changing ONLY the csv file as a starting point. That said, I don't understand why changing only one csv file and nothing else results in so many of the CI tests failing. Is that expected? I just cloned, updated the csv and pushed to this branch. Feel free to commit more changes or let me know if you have specific requests.

I can update update_column_descriptions and fix the tests. One set of tests involves some stringent QA on the CSV file itself, so it's not surprising that adding a new column would require updating the tests.

Posting what I put on Slack with regards to DESINAME:

I might suggest for consideration:
“Name with truncated decimal TARGET_{RA,DEC}”

I am not opposed to Stephanie’s version but it loses some important information while repeating some things that could be gleaned from the names themselves. If someone doesn’t explicitly look at a DESINAME entry then my explanation is less obvious, but if they do look at the description + entry then I think mine conveys more. So it depends on the most common use case we expect to have.

I’m very happy with the long description.

coverage: 100.0%. remained the same
when pulling 547accc on 195-column_descriptions-with-shorter-descriptions-for-fits-headers
into 6937e98 on main.

Unfortunately desiutil/bin/annotate_fits takes the first column with case-insensitive "description" anywhere in the name, not just the column called case-insensitive "description" as we had previously expected. That means that annotate_fits is picking up the "FullDescription" column instead of the intended "Description" column. Options:

• Reorder column_descriptions.csv to have "Name,Type,Units,Description,FullDescription" instead of "Name,Type,Units,FullDescription,Description"
• Rename "FullDescription" to something that doesn't have "description" as a subset of the name
• Update desiutil.annotate.find_column_name so that the prefix parameter actually mean prefix, and not "contained in", e.g. changing line 52 from if s in column.lower() to if column.lower().startswith(s). I don't know if there are other places that might be replying upon the current behavior, e.g. using a file with a column name "ColumnDescription".

@weaverba137 @stephjuneau opinions?

Again, my preference would be to not modify desiutil. Any solution that does not involve that is fine with me.

• I updated column_descriptions.csv to use the order "Name,Type,Units,Description,FullDescription". This lets desiutil/bin/annotate_fits work with the short Description column without requiring any code changes there.
• I updated the desidatamodel.stub code to use FullDescription instead of Description when updating a datamodel file.
• I updated the short version of DESINAME to Anthony's suggestion "Name with truncated decimal TARGET_{RA,DEC}" i.e. emphasizing the truncated aspect that would otherwise be tricky to know.
• I cross checked the new column_descriptions.csv FullDescription with the current main Description. I fixed one quote typo induced by the csv -> spreadsheet -> csv round trip, and verified that the other changes are purposeful:
  • DESINAME: more detailed FullDescription than before
  • SHAPE*_*: fixed ellipticity typo (thanks Stephanie!)
  • SURVEY: more detailed FullDescription than before