Hi Nicolas On Mon, Mar 02, 2026 at 01:48:51PM +0100, Nicolas George via ffmpeg-devel wrote:
[...]
> That leads me to the second part of this mail: practical questions about
> using Forgejo purely in command-line.
>
> It is something that was promised by people in favor of using this tool:
> that everything we could do in command-line, we would still be able to
> do in command-line, and that they would document how to do it. Time has
> passed, and that promise has still not been kept.
>
> So, proponents of Forgejo, please answer these questions:
First id like to clarify, iam not a proponent of web apps, iam replying
as i might have some secret forbidden knowledge.
some command line ways to access forgejo:
Listing pull requests:
gcli -a ff pulls -A michaelni
NUMBER STATE MERGED CREATOR NOTES TITLE
22301 open no michaelni 4 Suggest 0/0 explicitly as unknown
aspect ratio
22292 open no michaelni 1 avformat: Add
ff_format_check_set_url() and use in rtsp
22287 open no michaelni 0 avcodec/notchlc: lz4_decompress()
avoid byte loops, try to work in blocks
...
Listing comments:
gcli -a ff pulls -i 22301 comments
AUTHOR : jamrial
DATE : 2026-Feb-27 01:58:23
ID : 28628
Personally I think we should make it explicit that 0/1 is
unknown/undefined
and not 0/0, because the latter can result in divisions by zero when
used as
an AVRational. If no default is defined, every initialization of
AVRationals
(Be it by AVOptions or otherwise) sets them to it for that reason
after all.
0/0 should be considered an invalid value, so the docs should either
state
that "0" is used to refer to 0/1, or just replace all instances of "0"
with
it.
AUTHOR : GyanD
DATE : 2026-Feb-27 11:57:41
ID : 28680
Yes, keeping 0/1 for unknown allows tests to only check the numerator.
AUTHOR : michaelni
DATE : 2026-Feb-27 19:40:51
ID : 28833
@jamrial wrote in
https://code.ffmpeg.org/FFmpeg/FFmpeg/pulls/22301#issuecomment-28628:
| Personally I think we should make it explicit that 0/1 is
| unknown/undefined and not 0/0,
This is not how mathematics work. And not even an unanimous vote will
have
even a small effect on that. 0 is defined because it’s just a normal
number
in our number system: the additive identity.
the aspect ratio is width / height if you make 0 undefined you have
bugs and
require many special cases simply because it is mathematically wrong
lets just go over all cases to show this turning the image by 90° or
transpose is aspect = 1/old_aspect math: 1/(0/0) = 0/0 but yours:
1/(0/1) =
1/0 ( this is no longer what you said is undefined)
But thats not an exception lets compare 2 aspect ratio aspect1 <
aspect2 0/0
- a/b
(0b - a0)/(0*b)
0/0 you see it again become undefined, its the correct awnser as a
undefined
aspect compared to a defined one is undefined
but if you make it 0 0 - a/b -a/b thats a wrong awnser so you need to
add a
special code for every comparisson
you can try this with other operations, you will see 0/0 works 0
doesnt as
undefined
| because the latter can result in divisions by zero when used as an
| AVRational.
it should not, but you can open a bug report and assign it to me if it
does.
AVRational fully supports 0/0 and infinite
>
> 1. How do we get a notification for every comment made on any pull
> request or issue, including threading information about who answers
> to whom?
Iam not sure that there is any "threading" information beyond quotes in comments
>
> 2. How do we get notifications for actions done on the project,
> especially approving or closing a pull request?
I have a seperate michaelni-watch account and it seems to get mails for
everything,
I think it was simply subscribed to the ffmpeg project
>
> 3. How do we reply to one of the comments?
gcli -a ff comment -p22301 -R 28628
this opens a editor, i guess the one configured by EDITOR
just tried this and it works
> 4. How do we close a pull request?
gcli -a ff pulls -i 22301 close
should work, i have not tried
>
> 5. How do we approve a pull request to have it merged?
gcli -a ff pulls -i 22301 approve
gcli -a ff pulls -i 22301 merge
should work, i have not tried
thx
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
The worst form of inequality is to try to make unequal things equal.
-- Aristotle
signature.asc
Description: PGP signature
_______________________________________________ ffmpeg-devel mailing list -- [email protected] To unsubscribe send an email to [email protected]
