This notebook fails the oldestdeps job due to lack of .section support in astropy 5.3.0. As written, the notebook requires astropy>=7.0.2 (specifically, astropy/astropy#17611). The best way I see right now to make it compatible with v5.3.0 is shown in the diff below. The first two (replace .section with .data) are the most problematic because they load more data, which causes the notebook to take more than 2x longer to run (~6 min vs ~14 min). It also causes at least one figure to be different (screenshot below). I'm sure we could fix the figure but I don't see a way around the longer runtime. So rather than forcing the longer method on everyone just to support old astropy versions, I'm going to make the notebook require astropy>=7.0.2 and exclude it from the oldestdeps job.
$ diff MOST_queries.md MOST_queries-oldestdeps.md
610c610
< cutout = Cutout2D(hdul[0].section, position=cutout_center,
---
> cutout = Cutout2D(hdul[0].data, position=cutout_center,
661c661
< ref_cutout = Cutout2D(hdul[0].section, position=cutout_center,size=2*cutout_size,wcs=img_wcs,
---
> ref_cutout = Cutout2D(hdul[0].data, position=cutout_center,size=2*cutout_size,wcs=img_wcs,
720c720
< prev_scanID = ordered_metadata['scan_id'][0]
---
> prev_scanID = str(ordered_metadata['scan_id'][0])
728c728
< scanID = ordered_metadata['scan_id'][counter]
---
> scanID = str(ordered_metadata['scan_id'][counter])
864c860
< prev_scanID = ordered_metadata['scan_id'][0]
---
> prev_scanID = str(ordered_metadata['scan_id'][0])
872c868
< scanID = ordered_metadata['scan_id'][counter]
---
> scanID = str(ordered_metadata['scan_id'][counter])
Astropy v8.0.0 with .section on left, v5.3 with .data on right.
Originally posted by @troyraen in #337 (comment)
This notebook fails the oldestdeps job due to lack of
.sectionsupport in astropy 5.3.0. As written, the notebook requires astropy>=7.0.2 (specifically, astropy/astropy#17611). The best way I see right now to make it compatible with v5.3.0 is shown in the diff below. The first two (replace.sectionwith.data) are the most problematic because they load more data, which causes the notebook to take more than 2x longer to run (~6 min vs ~14 min). It also causes at least one figure to be different (screenshot below). I'm sure we could fix the figure but I don't see a way around the longer runtime. So rather than forcing the longer method on everyone just to support old astropy versions, I'm going to make the notebook require astropy>=7.0.2 and exclude it from the oldestdeps job.Astropy v8.0.0 with
.sectionon left, v5.3 with.dataon right.Originally posted by @troyraen in #337 (comment)