Front page | perl.perl5.porters |
Postings from July 2010
Re: backporting into 5.12.2 -- was kill $1 broken
Thread Previous
|
Thread Next
From:
Nicholas Clark
Date:
July 4, 2010 09:45
Subject:
Re: backporting into 5.12.2 -- was kill $1 broken
Message ID:
20100704164547.GM31795@plum.flirble.org
On Sat, Jul 03, 2010 at 05:38:03PM -0400, Jesse Vincent wrote:
>
> > However, I didn't think the expansion of commit bits was intended to
> > get more people applying other's patches. I thought it was to let
> > dual-life maintainers update their modules directly without burdening
> > the existing committers. Then some few of them (like me) have gone on
> > to apply patches.
>
> I can't speak for the pre-me policy, but getting in some more fresh
> blood to help with application of other contributors' patches is
> certainly part of my intent :)
My intent, and what I wrote in the e-mail I sent to various people encouraging
them (begging them) to accept was that they had contributed various good
patches that had been applied without needing any tweaking or review, hence
if the *only* thing they ever did was apply their own patches in future, that
would still be very helpful.
Of course, since then, for whatever reason, most haven't written another patch.
Maybe they'd actually finished fixing all the bugs they were comfortable with
fixing.
(Yes, for some other people the clear request was that they help by keeping
dual life modules up to date. But my view, and I speak only for myself, is
that we don't need all dual life module authors having direct commit access
provided that sufficient people are prepared to volunteer to do all modules
and hence spread the load. Which seems to be happening.)
To my mind the qualifications to having core commit access are being
competent to know your competencies, and having the right sort of diplomacy
skills to deal elegantly with bad patches and proposals
On Sat, Jul 03, 2010 at 06:57:25PM -0500, Craig A. Berry wrote:
> On Sat, Jul 3, 2010 at 10:00 AM, David Golden <xdaveg@gmail.com> wrote:
> For a number of reasons I'd be cautious about comparing 20 years of
> pre-git history with a year and a half of post-git, but I think you're
> right that the second and third tiers don't drop off quite as
> precipitously as before. I suspect that has more to do with jesse's
> recruitment efforts than with git, but every little bit helps.
Yes, my suspicion too. I doubt that we have enough years to prove it yet.
If it is the case, we should see a correlation between numbers of commits
and numbers of releases made, once some keen people have done their second
or third.
> The problem is that while a dozen or two new people contributing a
> dozen or two patches a year is a good thing, it doesn't substantially
> address the burden/risk of one or two people authoring 500-1000
> patches a year and no one else even coming close (and in the days of
> Jarkko that was one person authoring 1200-2500 patches a year).
Yes, and having the in depth knowledge that gives them.
> > I didn't think the expansion of commit bits was intended to
> > get more people applying other's patches. I thought it was to let
> > dual-life maintainers update their modules directly without burdening
> > the existing committers. Then some few of them (like me) have gone on
> > to apply patches.
>
> I think there are as many reasons for a commit bit as there are
> committers. I didn't mean to imply that anyone only pushing their own
> patches is doing something wrong, just that we're not keeping up and
> more committers hasn't helped much.
Agree.
And there are tasks that can be done by anyone, without a commit bit,
if they have the time. Specifically:
1: reading and attempting to replicate incoming bug reports. If it's a
regression, running a bisect to find out when it happened
2: Reviewing pure Perl code
3: If you're feeling keen - attempting to apply patches sent, and running make
test. Some people's patches don't even apply. Some don't pass tests. This
does happen.
More often, some people's patches will have passed on their machine/OS
setup, but some "random" person trying it on their machine may spot
something
Yes, most people won't have the time to do the above. But everyone?
Nicholas Clark
Thread Previous
|
Thread Next