Re: [gentoo-proxy-maint] [RFC] New doc & policy

[Michał Górny]

On czw, 2017-07-20 at 10:33 +0200, Kristian Fiskerstrand wrote:
> Thanks for doing this work Michał,
>
> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > Please note that we reserve the right to reject new packages, especially
> > if they would otherwise qualify for removal per our quality standards:
>
> Should we add something about the usefulness of packages for other
> users? As stated earlier, by being in main tree (and in particular part
> of stable set) there is support form other projects, any new package
> addition adds overhead.

I was thinking about that but I couldn't figure out a sane way to word
it. After all, I don't want to sound like we will judge every submission
for its usefulness to Gentoo users. And I certainly don't want to sound
like we're going to reject packages if e.g. they apply only to some rare
hardware.

I agree with your point but I'm just not sure if we should explicitly
mention it. After all, it still falls into 'rejecting at our own
discretion'.

> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > ===Taking over an existing package===
> > As a proxied maintainer, you can also (co-)maintain existing Gentoo
> > packages (both those with no maintainer, maintained by other proxied
> > maintainers and maintained by Gentoo developers/projects). In order to
> > take over the maintenance of a package, submit a commit adding yourself
> > to the {{Path|metadata.xml}} files. Usually, this commit will be
> > included along with other changes to the package.
>
> In the case of already maintained packages, they likely should
> communicate with the original maintainers first. Although I see you're
> touching upon it in next paragraph, the workflow seems to be
> communication after having worked up a patch and submitted it?

I will think how to improve this. Basically the intent was to emphasize
that we don't want people who just claim to maintain stuff; we want
people who are actually going to do it, and they are supposed to prove
it via fixing some issue.

Strictly speaking, it would probably be better to talk to the current
maintainer first. However, I would like to avoid a situation when
the other maintainer says 'fine', then the proxied maintainer demands
being added because of maintainer approval without actually proving
anything.

> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > There are a few points to consider:
> > * If the package has an existing maintainer, he will be asked to ACK
> > your changes and approve your request. Gentoo developers can reject a
> > proxied maintainer if they have a good reason to do so.
>
> They don't really need good reasons

Yes, they do. I'm against Gentoo where maintainers just reject everyone
because they like working alone. Think of the games team.

> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > ===GitHub pull requests===
> > The preferred way of submitting your contributions is through [https://g
> > ithub.com/gentoo/gentoo/pulls pull requests on our GitHub repository].
>
> Preferred by whom? (The preferred way to submit patches to gentoo
> proxied maintainer project...?)

This is proxy maintainers project documentation, so yes, it's about the
proxy maintainers team. If you don't want 'preferred', please suggest
a better word. The fact is, pull requests are handled more timely than
bugs (think of all the bugs that are still waiting for someone to take 
a look at them), and have CI coverage.

> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > ===Bug reports===
> > The classical method of submitting your changes is to attach them to the
>
> Would likely flip the order, since this is the classical way

Flipping the order would mean people read this first. This would mean
they're more likely to actually do it this way. Which in turn would mean
they will hit sheer frustration when someone asks them to file a pull
request instead if they don't want to wait months for someone to take
a look at it.

> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > ==Being a proxied maintainer==
>
> Should we add something on proper commit messages? And atomic changes
> per commit etc.

This really belongs in generic Gentoo documentation. I think linking
the 'Gentoo GitHub' page as the most current policy reference would
help.

>
> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > The skipped profiles include:
>
> ... profiles at the time of writing include..

Done.

>
> On 07/20/2017 10:02 AM, Michał Górny wrote:
> > ===Proxied maintainer in metadata.xml===
>
> mention of bugzilla account requirement?

Done.

-- 
Best regards,
Michał Górny