[3] ~~Gitlab/Github integration~~

[Florian Weimer]

* DJ Delorie:

> "Andreas K. Huettel" <dilfridge@gentoo.org> writes:
>> There's nothing inherently wrong with a workflow based on mailing lists
>> and patches,
>
> There's nothing inherently wrong with a workflow based on paper tape and
> in-person code reviews.  There's nothing wrong with a workflow based on
> modems and BBSs.  There's nothing wrong with a workflow based on gitlab
> PRs and web forums.
>
> Saying "there's nothing wrong with what we're doing now" misses the
> point - we're the only ones still doing it.  If we want to attract more
> new contributors, we have to go where the new contributors are.

Patch submission by mail is sufficiently flexible (especially with DC)
that it technically does not block anyone from collecting patches and
pre-reviewing them somewhere else, and then submitting them to
libc-alpha for final upstream review (perhaps with a From: line in the
body and a double Signed-of-by:).  It would be quite similar to what we
already do for people who do not feel comfortable posting to libc-alpha
directly.  (One odd thing about the third-party Github mirror discussion
at Cauldron was that it was predominantly focused on preventing
contributions.)

I do have concerns about long-term viability of email-based workflows
because the future of corporate email looks a bit bleak.  Especially
those of us who didn't standardize on Lotus Notes decades ago and lack
the institutional knowledge that developers need something else for
upstream work.  But for now things still work well enough.  I don't
think we would need months to make a switch to a forge workflow in case
email becomes increasingly unworkable, so I don't think it's necessary
to prepare for that possibility today.  Especially since we could use a
temporary stop-gap solution on the free/low-cost public forges.

Thanks,
Florian