Have a roadmap to re-review existing affiliated packages

[pllim]

We have collected a number of affiliated packages over the years (42 at the time of writing this). To ensure that the quality of being an affiliated package is maintained, there is a need to go back and re-review them.

Possible outcome for an existing affiliated package:

• Stay affiliated: If badges remain the same or have improved overall.
• Get unaffiliated (after a grace period for the package to try to fix problems): If the package is outright abandoned, or badges have worsened overall, or any of the badges turns red.

cc @hamogu and @astropy/coordinators

[astrofrog]

I agree with this!

[pllim]

Sorted the existing affiliated packaged (not coordinated) by last updated. Is this a good place to start, by going down this list?

name	repo_url	last_updated
Saba	https://github.com/astropy/saba	2017-09-26
galpy	https://github.com/jobovy/galpy	2017-11-08
synphot	https://github.com/spacetelescope/synphot_refactor	2017-11-08
hips	https://github.com/hipspy/hips	2017-11-08
poliastro	https://github.com/poliastro/poliastro	2017-11-08
linetools	https://github.com/linetools/linetools	2017-11-08
pyspeckit	https://github.com/pyspeckit/pyspeckit	2017-11-08
marxs	https://github.com/Chandra-MARX/marxs	2017-11-08
cluster-lensing	https://github.com/jesford/cluster-lensing	2017-11-08
Halotools	https://github.com/astropy/halotools	2017-11-08
stingray	https://github.com/StingraySoftware/stingray	2017-11-08
HENDRICS	https://github.com/StingraySoftware/HENDRICS	2017-11-08
gala	https://github.com/adrn/gala	2017-11-08
naima	https://github.com/zblz/naima	2017-11-08
Astro-SCRAPPY	https://github.com/astropy/astroscrappy	2017-11-08
omnifit	https://github.com/RiceMunk/omnifit	2017-11-08
spectral-cube	https://github.com/radio-astro-tools/spectral-cube	2017-11-08
ginga	https://github.com/ejeschke/ginga	2017-11-08
PyDL	https://github.com/weaverba137/pydl	2017-11-08
APLpy	https://github.com/aplpy/aplpy	2017-11-08
astroplan	https://github.com/astropy/astroplan	2017-11-08
gammapy	https://github.com/gammapy/gammapy	2017-11-08
Glue	https://github.com/glue-viz/glue	2017-11-08
sncosmo	https://github.com/sncosmo/sncosmo	2017-11-08
imexam	https://github.com/spacetelescope/imexam	2017-11-08
spherical_geometry	https://github.com/spacetelescope/spherical_geometry	2017-11-08
gwcs	https://github.com/spacetelescope/gwcs	2017-11-08
pyregion	https://github.com/astropy/pyregion	2017-11-08
dust_extinction	https://github.com/karllark/dust_extinction	2018-05-25
corral	https://github.com/toros-astro/corral	2018-08-21
feets	https://github.com/carpyncho/feets	2018-11-27
baseband	https://github.com/mhvk/baseband	2018-11-27
sbpy	https://github.com/NASA-Planetary-Science/sbpy	2019-06-18
EinsteinPy	https://github.com/einsteinpy/einsteinpy	2019-07-23
astroalign	https://github.com/toros-astro/astroalign	2019-08-15
astroML	https://github.com/astroML/astroML	2019-10-06
PyVO	https://github.com/astropy/pyvo	2019-10-06
mocpy	https://github.com/cds-astro/mocpy	2019-11-04
ligo.skymap	https://git.ligo.org/lscsoft/ligo.skymap	2020-07-01
agnpy	https://github.com/cosimoNigro/agnpy	2020-07-13
lenstronomy	https://github.com/sibirrer/lenstronomy	2020-07-24
statmorph	https://github.com/vrodgom/statmorph	2021-01-12

p.s. Found a bug! astropy/astropy#10823

Update: Changed statmorph date due to developer-initiated re-review (astropy/astropy.github.com#421)

[adrn]

What does "last updated" mean?

[pllim]

Not sure... Maybe CoCo knows?

[adrn]

I only ask because it's definitely not the date that each package was last updated (gala has a release from July 2020 and commits from days ago), so it must be a log of when the affiliated package record was last updated..?

[adrn]

Oh yea, they seem to be static values from the affiliated package JSON file - sorry for the noise!

[pllim]

Looks like @bsipocz explained it at https://github.com/astropy/astropy.github.com/pull/382/files#r448593260

I think this date is for the review status and relevant to see when the package needs a rereview, etc.

[astrofrog]

Yes just confirming it is when the review results were last updated. We could rename the key to include review perhaps?

[eteq]

I also agree with doing this and confirm that that date was intended for exactly the purpose here (to know when to re-review).

That said... I feel like there's an argument for trying to hold the number at 42... 😉

[pllim]

@hamogu , how about we plan to start the cycle, say Jan 2021, and strive to re-review one package every 2 months? (That will only take us 7 years...)

[astrofrog]

7 years assuming no new packages!

[pllim]

We could try parallel processing but will need more cores. 🤓

[astrofrog]

Hopefully would work better than the io.ascii parallel reader?

[pllim]

Also see #334 . I think we're leaning towards automation and not a full re-review.