Hi,

Addressing your concerns individually,

On Sat, Mar 21, 2026, at 4:51 PM, Michael Niedermayer via ffmpeg-devel wrote:
> First, a dependancy on external infrastructure is unnecceasry

The PR in question does not add a dependency on any external infrastructure. It 
actually does not add any external dependency, precisely because it is a 
vendored inclusion rather than an external reference.

Actually, my own plan would have been to allow linking against an externally 
provided libcheckasm (instead of ffmpeg's own checkasm). The idea to vendor 
checkasm and integrate it into FFmpeg's build system at all was Martin's idea.

(And to be fair, vendoring it will hopefully prevent or slow the rate at which 
they would inevitably begin diverting again; and perhaps more importantly, will 
allow checkasm tests to start depending on new libcheckasm APIs)

> Second, a dependancy on gitlab-CI is not acceptable
> We need to be able to run the full relevant test coverage ourselves, 
> under FFmpeg-controlled infrastructure

I think this may have been a simple misunderstanding? This PR does not add a 
dependency on any external CI to run checkasm tests. All of FFmpeg's checkasm 
tests are still defined locally, and run locally, inside FFmpeg's own CI.

The references to external CI, as was pointed out by Martin, are for checkasm's 
*own* self-tests, useful only to developers of libcheckasm itself. Something 
that FFmpeg's checkasm does not currently have at all, so even with those 
self-checks removed, the status quo gets no worse.

> Third, the copyright and authorship situation needs clarification.

I think this was sufficiently addressed by Martin. Actually, if you are worried 
about transparent copyright/authorship information, that would be a point in 
favor of a merge (or submodule), rather than a squashed code dump.

> Forth,
>     "Some changes have been ported over from FFmpeg's copy of checkasm, 
> with permission to relicense."

This refers to commits e0d56f097f42bcdbe6c3b2f57df62a4da63f2094 by Clément 
Bœsch, and commit 711781d7a1714ea4eb0217eb1ba04811978c43d1 by Janne Grunau. I 
have written permission to use both commits under the terms of the BSD-2 clause 
license, by the relevant authors.

Apart from that, the only other code that was ported from the FFmpeg checkasm 
was Rémi's RISC-V harness, which he has contributed to libecheckasm by himself.

>     My permission for relicensing was conditional on the resulting 
> project being hosted on code.ffmpeg.org.
>     That condition was never fulfilled. So unless I am missing 
> something, that statement is at best
>     incomplete and may be misleading.

Can you clarify which exact commit of yours do you think that libcheckasm 
infringing the copyright of?

>
> Fifth,
>     the libcheckasm work was done by FFlabs people and discussed in 
> FFlabs meetings.

I don't see any FFlabs member in the list of authors you listed above. As you 
are no doubt aware, the bulk of the work was authored by myself and Martin, 
neither of whom, to my knowledge, is employed by FFlabs.

>     So unless I misunderstand the situation, we seem to have funded 
> work that now removes FFmpeg attribution or identity,

When you say "we", do you refer to the FFmpeg project, or the private 
consulting company you are a co-founder of?

> Please correct me if I am wrong, but from where I stand this proposal 
> raises significant technical,
> governance, infrastructure, and attribution concerns, and I do not 
> think it should be merged in its current form.

You have raised infrastructure and attribution concerns. What technical 
concerns did I miss?
_______________________________________________
ffmpeg-devel mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to