[FFmpeg-devel] Notifications for the new monolithic tool

[Timo Rothenpieler]

On 7/24/2025 6:48 PM, Nicolas George wrote:
> Timo Rothenpieler (HE12025-07-24):
>> I don't see any need for this, no.
> 
> If the patch does not reach our inbox, it will not be reviewed.
> 
>> That's what it does, just not as format-patch, but simply as a link.
> 
> A link is not enough. When we open our mail, we need to be able to see
> the proposed change without an extra step.

Why?
And who are "we"?
Having a working Web-browser is a reasonable requirement in 2025.

One could even make the counter argument that only being able to see 
patches as mail is not enough, and actively deters contributors and 
potential future maintainers.

>> And the mailing list is "a monolithic thing" as well, so there's no point
>> throwing shade like that.
> 
> The mailing-list is just a mailing list, it does not try to combine a
> dozen different tools into one.
> 
>> Patchwork tries to do that and it only semi-works, so I'd rather not.
> 
> Then stop trying to force us to use mediocre tools.

The way it looks to me, there is only you, not an "us" or "we".
I have not seen anyone except for you oppose the use of more modern tools.
The only discussion ever was about which ones exactly, and hosted where 
by whom.

The general expectation even seems to be that the ML will eventually 
become obsolete, though I don't see that happen anytime soon.

>>> # View it on FFmpeg Forgejo ( https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/20031 ) or reply to this email directly.
>>> is not adequate.
>> Yes it is, the mailing list just eats the Reply-To address, breaking it.
> 
> No it is not: it requires an extra step.
> 
>> It's been mentioned multiple times on the list and on IRC by now, and it
>> should be obvious that you can't just give yourself push/merge access on
>> your own anyway.
> 
> I do not care about push access for now, I care about notifications.
> 
> As libabvfilter maintainer, if somebody proposes a patch on libavfilter,
> I need to see it.
> 
> You promised it would be possible to continue working through mail
> without extra work: did you lie?

When did I promise that?
That was something that was discussed at some point I think, but I do 
not remember _anyone_ making such a promise.

It does not seem like a realistic goal to me either. Translating mails 
to patches is not something that's easily done, as proven by patchwork.

Once the spamminess problem has been figured out, notifications about 
PRs will be forwarded to the ML again. So that direction will work fine, 
just not as format-patch patches, but references to PRs.

What I do remember is the requirement to be able to largely/completely 
interact with the tool via CLI.
And you can interact with forgejo nearly entirely via cli with one of 
multiple tools made for it (and plain git itself of course).
The most feature complete being https://gitea.com/gitea/tea
There's multiple ones in active development, so you can pick your favourite.

And you can reply to notification mails from it directly via mail, just 
not when the ML is forwarding them it seems.