Re: [gentoo-dev] Is there any way I can help with pull requests?

[Ralph Seichter]

On 16.10.18 20:25, Virgil Dupras wrote:

> Maybe I can try to explain why your 3 PRs [1] are still opened.

Thank you, I appreciate that, but let me repeat that my question about
helping with PRs was not meant as criticism.

> The "skel.ebuild" one is easy: global changes have to be discussed on
> gentoo-dev. [...]

You'll find that I only mentioned two open PRs of mine, because I took
the skel.ebuild PR deliberately out of consideration. The conversation
in that PR made it clear that it is a larger issue, to be handled by QA.
No surprise there, let it simmer. ;-)

> The milter-regex one is, I think, a result of miscommunicating intent.
> [...] Someone from that project [2] is going to have merge it, not
> zlogene.

Ah, that's news to me. Should the PR then be assigned to the Net-Mail
project in a publicly visible way? I'd like to see this merged, because
it introduces a new milter-regex feature, one which I asked for and the
author was kind enough to implement (and I also helped him to test it).

> it's completely understandable that you expect a timely response to
> your correction, but ultimately, you'll have to nudge someone from
> the net-mail project.

I had no idea that it is my responsibility to move this PR along. I had
naively assumed that once a Proxy Maintainer project member had reviewed
it, as it happened here, the process would continue without me unless
more changes were asked for later on. Can you please tell me how I best
hand this pull request to the Net-Mail team?

> Then, we're left with your nginx-unit PR, which is part of the
> proxy-maint program. In this case, we have mgorny who doesn't
> seem to like your PR.

I followed the PHP team's recommendation, as can be seen from the PR
conversation and from the underlying Bugzilla report. While I respect
Michał Górny's opinion, I understand that he does not have a deciding
vote in this case. Michał may not fully agree with what the PHP team
recommended, but I was told he's in the Python team. Let me quote from
a conversation with Michael Orlitzky here (I have permission):

  <Orlitzky>
  [Michał is] probably just busy. His last comment didn't sound like he
  was strongly against it.

  He's probably thinking of it in terms of python (he's on the python
  team), where things are set up a bit better. With python stuff,
  PYTHON_TARGETS says what versions of python you want to use with the
  thing you're building. For example, you would probably use
  PYTHON_TARGETS if you wanted to support multiple python versions in
  nginx-unit. In that case, it's fine -- that's what PYTHON_TARGETS is
  for.

  We don't have a variable like that for PHP ebuilds. If you install
  some PHP code and switch to an interpreter that it doesn't work
  with... sorry, it'll just crash. Fixing that (like python/ruby do it)
  would be a huge effort and there's just not enough people interested
  in PHP. A long time ago, though, we needed a variable that let us
  build *extensions* for specific versions of PHP (this problem is a
  little easier to solve), aThe PHP team at the time stole the name
  *_TARGETS from python, even though it's not quite the same thing.
  Which brings us to why Michal is probably thinking PHP_TARGETS is what
  you want. It doesn't do the same thing as PYTHON_TARGETS, though.

  Feel free to quote me on any of this =)
  </Orlitzky>

Again, I understand and respect that not everybody has the same view on
things. I asked for the PHP team's advice, followed it, and if a third
party does not agree, it does not bother me much. That's why there are
different teams, after all. It is good that Michał Górny communicated
his concerns, but he's the only person who spoke up and had no better
alternative to offer. The modified ebuild (based on my own original)
works fine, and I'd be glad to see this moved along. There was a bug
filed for the lack of PHP support, and the PR also bumps the revision
to the latest production release, made approx. one month ago.

> As I hope to have demonstrated, there is no ill intent or even
> negligence in the result that you observe.

I had not suspected negligence or malice, but I am grateful for your
explanations. I learned more about the process today. Thank you, Virgil.

-Ralph