Links to old PyFITS issues in regression tests

[MSeifert04]

There are several links to old PyFITS issues in the test suite of io.fits. @pllim pointed out (#6685 (comment)) that some of them might not be around forever. I'm not sure if that's true for the GitHub repo but definitely for the TRAC links.

So it could make sense to migrate them somewhere else where "we" control the lifetime?!

[MSeifert04]

There are 218 issues (total) on trac, with 65 being mentioned in the code. That's a lot of issues.

I'm not sure if we want 65 (or 218) migrated issues here...

[bsipocz]

You mean 65 open issues that are mentioned in code?

[pllim]

Wow maybe that's not feasible. @embray did do a issues migration with https://github.com/spacetelescope/github-issues-import but is it worth the effort or even applicable here?

For now the Trac pages are still accessible but read-only. I'll inform you if I hear otherwise.

[MSeifert04]

@bsipocz No. All of them are links to fixed bugs (the 65) - mostly in regression tests but some also in the actual code. All in all there were 218 issues on trac but we only link to 65 of them in the code.

[MSeifert04]

One other option would be to create a new package in the astropy-organization to just hold the migrated issues. Not sure if it's worth it...

[bsipocz]

Well, if we really want to copy over everything, there may be tools to do that, probably either https://github.com/trustmaster/trac2github or https://github.com/XhmikosR/migrate-trac-issues-to-github would be sufficient.

We just need to make sure it doesn't try to move over everything, only the issues and comments, and leaves labels, milestones and assignment alone.

[saimn]

Not sure if it is worth it, some issues are already fixed, others are already opened here or on https://github.com/spacetelescope/pyfits/issues. But to keep track of the history of these issues the simplest solution could to migrate all trac issues to a dedicated github repo with something like trac2github.

[astrofrog]

Another option would be to copy the content of the issues into a file and just include that file inside astropy/io/fits.

In general I wonder if this means that ideally we shouldn't link to issues but describe them sufficiently well in the test so that linking isn't needed?

[MSeifert04]

In general I wonder if this means that ideally we shouldn't link to issues but describe them sufficiently well in the test so that linking isn't needed?

In some cases the discussion for an issue could be very interesting. Especially if the fix was more a hack "to make it work" instead of "good" solution.

But in general: I agree, linking to an issue instead of describing the test should be a no-no.

[pllim]

I am tempted to just label this as "won't fix" for now and close it. We can re-open if STScI decides to remove those Trac links (although it is unlikely in the near future).

[embray]

Well that Trac no longer exists...(?) Is it archived somewhere?

[pllim]

Is it archived somewhere?

Probably in the IT dept cold storage... Not sure if they even remember where they put it. So I think we effectively lost the record. If you must find out, please submit a ticket to https://stsci.service-now.com/stars/

[astrofrog]

I would be in favour of just removing broken links to the trac