Re: FFmpeg as SPI associated project

In data Thursday 2012-05-31 13:33:34 +0100, Ian Jackson ha scritto:
> Stefano Sabatini writes ("Re: FFmpeg as SPI associated project"):
[...]
> This part of these project acceptance resolutions is not some kind of
> template which needs to just have the project name and the liason name
> plugged into it. This clause is supposed to state what SPI's
> understanding is of FFmpeg's governance structure. It is obviously
> important for SPI to know what FFmpeg's governance structure is so
> that SPI can honour it. (And that needs to be sorted out in advance
> of any dispute arising, so just because things are all rosy now
> doesn't mean it's not important.) The purpose of putting this clause
> in these accession resolutions is to make a clear public statement,
> for review both by SPI and the proposed project, of what the common
> understanding is.
>
> So as I say since FFmpeg is not an autocracy, it is not correct to say
> that "Stefano Sabatini is the authoritative decisionmaker for FFmpeg".
> Rather, that is a false statement. Therefore "SPI recognises that
> Stefano is the authoritative decisionmaker" is also inappropriate.
> SPI should not "recognise that {false statement}".
>
> Hence my proposed alternative wording:
>
> Stefano Sabatini is recognised by SPI as the current liason for
> FFmpeg. SPI expects him to inform us of decisions relating to SPI
> made by the FFmpeg project, and we will honour his requests in
> accordance with the Framework for Associated Projects.
>
> However FFmpeg does not currently have a formal governance
> structure. Therefore in case of significant dispute, SPI will
> follow what appears to the SPI Board to be the rough consensus
> view of the FFmpeg project committers.
>
> Is this a proper understanding and reflection of FFmpeg's governance
> structure ?

Yes, seems reasonable. "In case of significant dispute" I suppose in
case there is no agreement between project maintainer and liaison but
one of them want to force a request (unlikely but not inconceivable),
in this case SPI should simply not validate the request.

Also the term "committers" is not really valid in case of FFmpeg given
our recent history, I believe the term "direct contributors" is more
suited to the current status of the project.

[...]
> That's entirely for FFmpeg to decide.
>
> But to make this workable, we need to have a process from SPI's point
> of view that relies on trusting the liason for day-to-day stuff.
>
> That is, the point of having the liason is not that the liason's word
> is definitive with respect to what SPI does. The point is that the
> liason is responsible for communicating decisions (however made) to
> SPI. So whenever some decision (eg a spending decision) is made by
> FFmpeg (collectively, in whatever way is established for and by
> FFmpeg), the liason communicates that to SPI by saying `we have
> decided that we would like to spend $X on Y' and the SPI Treasurer
> writes a cheque or whatever.
>
> If the liason breaches FFmpeg's processes (for example by sending
> unilateral requests to SPI for money to be spent, or whatever) they
> would, in effect, be lying to SPI.

In the application we provided a mechanism which allows to verify if a
fund request has been approved:

a fund request is discussed in the public mailing list, and approved
by the liaison; it is also approved by the project maintainer, or 7
days pass with no explicit disapproval

in this case the liaison will send a mail with a link to the public
mailing list with the request approval. The SPI treasurer can directly
verify that the request was approved.

> We hope that if that should happen, someone else from FFmpeg would
> notice and bring it to SPI's attention. Then SPI would consult the
> accession resolution to see how to resolve the dispute. So the
> accession resolution needs to say what to do.
>
> In this case Robert's wording would imply that the dispute would be
> resolved by doing whatever Stefano Sabatini says. (I know Robert
> doesn't seem to think that, but I think that's what `authoritative
> decisionmaker' means. Even Robert should see that we need a wording
> which is not subject to disputes about its meaning.)
>
> In this case my wording would mean someone from SPI would have to
> figure out who the current FFmpeg committers were, and ask them.
>
> Would that be correct ? If so, then we should IMO have my wording or
> something like it. If not then we need something different.

OK, so we need to understand what we exactly need to do in order to
comply with the SPI terms, and in particular how we need to advance in
terms of "formal governance structure" so SPI knows how to deal in
case a "significant dispute" arises.

Best.