Inconsistency between resolved issues/patches and source files.

Inconsistency between resolved issues/patches and source files.

[William]

Hello.

A few days ago I tried reencoding several video files I have laying
around to save storage, using libsvtav1 through ffmpeg, and then to my
surprise (and frustration) I realized that ffmpeg distributed by
GNU/Guix does not include such encoding library.

I checked the issue tracker and found out issues that indeed commented
on this library missing, strangely enough, an issue with a patch was
marked as solved on 16 February 2024[0], over a month ago, however when
I check the source files I notice that the patch that includes a flag to
enable this library is still not present on the latest build of ffmpeg
yet[1].


Why? How does the merging process work on Guix? Why are issues marked
as solved/done yet patches are still one month due, to submit upstream?
Am I missing something or this is a mistake on the mantainers side?

[0]: https://issues.guix.gnu.org/63095
[1]:
https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/video.scm#n1780

Re: Inconsistency between resolved issues/patches and source files.

[Vagrant Cascadian]

[-- Attachment #1: Type: text/plain, Size: 1270 bytes --]

On 2024-03-21, William wrote:
> A few days ago I tried reencoding several video files I have laying
> around to save storage, using libsvtav1 through ffmpeg, and then to my
> surprise (and frustration) I realized that ffmpeg distributed by
> GNU/Guix does not include such encoding library.
>
> I checked the issue tracker and found out issues that indeed commented
> on this library missing, strangely enough, an issue with a patch was
> marked as solved on 16 February 2024[0], over a month ago, however when
> I check the source files I notice that the patch that includes a flag to
> enable this library is still not present on the latest build of ffmpeg
> yet[1].

It appears to be applied in core-updates as
9400188b2b57cea2f3efc998a2f3b24769db04b7, but not yet merged into the
master branch. This is probably because it would trigger too many
rebuilds...


> Why? How does the merging process work on Guix? Why are issues marked
> as solved/done yet patches are still one month due, to submit upstream?
> Am I missing something or this is a mistake on the mantainers side?

When closing an issue, ideally such information (commit hash and branch
if not on master) should be included in the closing message to avoid
having to look it up. :)


live well,
  vagrant

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

[bug#63095] Inconsistency between resolved issues/patches and source files.

[Vagrant Cascadian]

[-- Attachment #1: Type: text/plain, Size: 1270 bytes --]

On 2024-03-21, William wrote:
> A few days ago I tried reencoding several video files I have laying
> around to save storage, using libsvtav1 through ffmpeg, and then to my
> surprise (and frustration) I realized that ffmpeg distributed by
> GNU/Guix does not include such encoding library.
>
> I checked the issue tracker and found out issues that indeed commented
> on this library missing, strangely enough, an issue with a patch was
> marked as solved on 16 February 2024[0], over a month ago, however when
> I check the source files I notice that the patch that includes a flag to
> enable this library is still not present on the latest build of ffmpeg
> yet[1].

It appears to be applied in core-updates as
9400188b2b57cea2f3efc998a2f3b24769db04b7, but not yet merged into the
master branch. This is probably because it would trigger too many
rebuilds...


> Why? How does the merging process work on Guix? Why are issues marked
> as solved/done yet patches are still one month due, to submit upstream?
> Am I missing something or this is a mistake on the mantainers side?

When closing an issue, ideally such information (commit hash and branch
if not on master) should be included in the closing message to avoid
having to look it up. :)


live well,
  vagrant

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 227 bytes --]

Re: Inconsistency between resolved issues/patches and source files.

[William]

On Thu, 21 Mar 2024 11:21:33 -0700
Vagrant Cascadian <vagrant@debian.org> wrote:

> On 2024-03-21, William wrote:
> > A few days ago I tried reencoding several video files I have laying
> > around to save storage, using libsvtav1 through ffmpeg, and then to
> > my surprise (and frustration) I realized that ffmpeg distributed by
> > GNU/Guix does not include such encoding library.
> >
> > I checked the issue tracker and found out issues that indeed
> > commented on this library missing, strangely enough, an issue with
> > a patch was marked as solved on 16 February 2024[0], over a month
> > ago, however when I check the source files I notice that the patch
> > that includes a flag to enable this library is still not present on
> > the latest build of ffmpeg yet[1].  
> 
> It appears to be applied in core-updates as
> 9400188b2b57cea2f3efc998a2f3b24769db04b7, but not yet merged into the
> master branch. This is probably because it would trigger too many
> rebuilds...
>
>
> live well,
>   vagrant

Hello.

Could someone explain exactly what the "core-updates" branch is?

Can perhaps "core-updates" be added as an extra channel on Guix
manually?

And how long usually it takes for these patches to make its way from
"core-updates" to master?

Thanks in advance.

Re: Inconsistency between resolved issues/patches and source files.

[Felix Lechner via]

Hi William,

On Tue, Apr 09 2024, William wrote:

> Could someone explain exactly what the "core-updates" branch is?

The mission of that branch is changing but it's not going away.
Originally, it pooled all edits that resulted in large numbers of
rebuilds on the substitute servers.

After team branches were allowed on Savannah, the branch now pools
updates to "core" packages. [1][2]

Personally, I think we should get rid of the branch. [3]

> Can perhaps "core-updates" be added as an extra channel on Guix
> manually?

You may be able to do that via your home or your system channels
configuration, but I haven't tried.  Instead, I maintain my own branch
which carries the modifications my systems require. [4]

I expect that situation to persist indefinitely.

> And how long usually it takes for these patches to make its way from
> "core-updates" to master?

On average about a year, or a year and a half.  That unusually long
period has been identified as an example where the project's performance
as a group falls short of our collective expectations.

Unfortunately, the fear of causing rebuilds on the substitute servers is
too dominant in our minds.  The group cannot muster the courage to
ignore it in the name of progress.

As nckx pointed out, "no one likes our Gentoo emulation mode."

Kind regards
Felix

[1] https://lists.gnu.org/archive/html/guix-devel/2023-12/msg00059.html
[2] https://lists.gnu.org/archive/html/guix-devel/2024-01/msg00096.html
[3] https://lists.gnu.org/archive/html/guix-devel/2024-02/msg00334.html
[4] https://codeberg.org/lechner/guix/src/branch/lechner-experimental