Trac to GitLab migration underway

[Ben Gamari]

Richard Eisenberg <rae at cs.brynmawr.edu> writes:

> Thanks to all who have made this happen. Looking forward to having
> everything in one place!
>
> Do we have any ability to edit the formatting of subject lines of
> ticket emails? It would be great if we could get the (!234) or
> (#12345) up front, as this is the effective summary of what's to
> follow.
>
I think we can manage that. Ideally I'd like to stay as close to an
unmodified GitLab installation as possible but this particular issue
doesn't sound like it should be a hard thing to fix.

> I'd also like to second Simon's request, in particular, of having a
> clear path from an issue to an MR. Getting from an MR to an issue is
> normally easier -- when the MR is created, most devs would
> automatically mention the issue number somewhere in the description --
> but going from an issue to an MR is harder. It would be a shame to
> have to search through the commentary looking for the link. I used the
> old "Phab:D1234" links in the top of a Trac ticket quite often.
>
In principle it shouldn't be any harder to go from an issue to an MR
than it is to go the other direction. For instance, consider the case of
#16347 [1]. You will note that below the "Related issues" section there
is a list of related merge requests (strangely formatted completely
differently).

To be honest, how this list is formed is a bit of a mystery to me. The
fact that !525 is included is not surprises: !525 mentions #16347 in
its description. However, !436 is a bit less obvious since it does not
mention #16347 at all. My hypothesis is that it is transitively included
via #16344, which does mention #16437 and is related to it.


[1] https://gitlab.haskell.org/ghc/ghc/issues/16347


> I also second the request for a test-case field. This field, to me, is
> most useful as a reminder to the author that they need to add a test
> case. But I think the issue-to-MR link is more important than the
> test-case field.
>
Indeed, as I mentioned earlier this may be a bit of a hard thing to
accomplish in the near-term. One measure we could take to ensure that
tests aren't forgotten is to include a "field" in the default merge
request description which contributors would be asked to fill in with
the names of the tests that cover their change. That being said, this is
no substitute for a proper field.

Cheers,

- Ben
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 487 bytes
Desc: not available
URL: <http://mail.haskell.org/pipermail/ghc-devs/attachments/20190311/fa1a4bc0/attachment.sig>