[FFmpeg-devel] [PATCH] MAINTAINERS: add a separate list for those with push access

Tue Feb 7 03:20:27 EET 2023

On Mon, Feb 06, 2023 at 01:10:06PM +0100, Lynne wrote:
> Jan 30, 2023, 20:03 by dev at lynne.ee:
> 
> > Jan 30, 2023, 17:49 by michael at niedermayer.cc:
> >
> >> On Thu, Dec 15, 2022 at 02:13:49AM +0100, Lynne wrote:
> >>
> >>> This list is incomplete, and just contains those I could see
> >>> while looking at the recent git log. If it looks like I've forgotten you, I definitely haven't!
> >>> We may complete the list at a later date.
> >>>
> >>> This makes it such that those who add themselves to MAINTAINERS do not
> >>> get push access by default, but rather, they have to request it
> >>> explicitly in a different commit. This used to be the situation
> >>> before it was changed at the start of this year and is pretty much what
> >>> everyone expects.
> >>>
> >>> Patch attached.
> >>>
> >>> MAINTAINERS |   15 +++++++++++++++
> >>> 1 file changed, 15 insertions(+)
> >>> 6a083061d75f6655771bde377f96aadad19b21c6  0001-MAINTAINERS-add-a-separate-list-for-those-with-push-.patch
> >>> From 5c353412a25fd46c5077e5cf92ddfd6532eb46cb Mon Sep 17 00:00:00 2001
> >>> From: Lynne <dev at lynne.ee>
> >>> Date: Thu, 15 Dec 2022 02:05:00 +0100
> >>> Subject: [PATCH] MAINTAINERS: add a separate list for those with push access
> >>>
> >>> This list is incomplete, and just contains those I could remember
> >>> while looking at the recent git log.
> >>> We may complete the list at a later date.
> >>>
> >>> This makes it such that those who add themselves to MAINTAINERS do not
> >>> get push access by default, but rather, they have to request it
> >>> explicitly in a different commit. This used to be the situation
> >>> before it was changed at the start of this year.
> >>>
> >>
> >> I dont object to you adding a list of people with commit acccess though i
> >> dont think its needed or that useful.
> >> But adding a list that is incomplete, sorted in a odd way and doing so in a
> >> commit that states a past rule which i dont think was true, seems not
> >> ideal
> >>
> >> ATM there are I think 117 keys that have write access (some may belong to
> >> the same developers) and also over 100 maintainers in that MAINTAINERs file
> >> I think. I didnt try to count them too precisely. But the numbers are not
> >> that disimilar. The added list is quite abit more different
> >>
> >
> > My intention was to make this complete after it's accepted (or not, if
> > someone doesn't want to be known for having push access).
> >
> >
> >> Also iam not sure this commit will change that much. People who do not want
> >> write access neither before nor afterwards will not send a ssh key so wont get
> >> write access. And people who want write access will push for it and
> >> probably noone will object. Theres the between people who dont push for
> >> it and noone else would push either they might no longer receive write
> >> access. Iam not sure if that is better.
> >>
> >> It makes things more involved but whats really bad is that this extra
> >> step is mainly in your mind, its not docuemnted.
> >> Do i add someone to that new list when i give him write access or do
> >> i give someone write access when a patch adding her is approved. Or do
> >> i just ignore that list because its incomplete anyway ?
> >>
> >> I assume the intend is the 2nd one but How would a contributor know
> >> to add herself to that list and what about people who are quite humble
> >> and who would not push for it yet at the same time would benefit from
> >> write access ?
> >>
> >
> > How would anyone know to maintain something they should add themselves
> > to the list of maintainers?
> > A second list of those with push access doesn't add more roadblocks, it's
> > just a separate list, that's all. You wouldn't have to add yourself to maintainers
> > to get push access if you don't want to.
> > As for those humble, I do see your point, but it's a one-line diff change,
> > and it can be done in the same commit adding yourself to maintainers,
> > it's not a 2-page personal statement about values.
> >
> >
> >> ATM every maintainer automatically receives the right for write access
> >> After this patch its made more difficult, i cant just post a patch adding
> >> random people either Someone would have to convince them first that they
> >> should post a patch to add themselfs. 
> >>
> >> So what i really dislike on this change is the potential stumbling blocks
> >> it throws before new developers.
> >>
> >> Its important that one has write access to the repository one works in
> >> In FFmpeg that work happens on git master so write access to that is
> >> important for anyone actively working on it.
> >> In other places work and review might happen in developers own repositories
> >> and they get merged regularly. In that case write access to master is not needed
> >>
> 
> At the FOSDEM meeting yesterday, everyone there agreed that while it's not
> perfect, it's a step in the right direction, and we should merge this.

Well, i was not there and i do not know what was said, also i dont think
the issues have been addressed

The commit message implies a past policy which is not correct and commit
messages can not be corrected later so it should be corrected first.

Also if the intend is that people need to add themselfs to the git list in
the MAINTAINERS file to get write access then this should be written in
the MAINTAINERS file and also somewhete in the policy, maybe the
patch checklist.
implied only by a commit message, i think it will be missed

Still i think this change is not a good idea. I think maintainership and
git access should be closely tied together. Not choosen one by one

Just to clarify, i dont mind at all to docuemnt properly who has git
access exactly (which may differ slightly) but the idea someone would be
a active maintainer and work / want to work but not be given git access
really feels like a bad idea to me. Also this whole makes the process more
complex

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

There will always be a question for which you do not know the correct answer.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 195 bytes
Desc: not available
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20230207/76980698/attachment.sig>