[FFmpeg-devel] [RFC] The Big Bump

[Reinhard Tartler]

On Fri, Feb 04, 2011 at 13:53:45 (CET), Anton Khirnov wrote:

> On Fri, Feb 04, 2011 at 09:36:06AM +0000, M?ns Rullg?rd wrote:
>> Stefano Sabatini <stefano.sabatini-lala at poste.it> writes:
>> 
>> > On date Thursday 2011-02-03 21:06:31 +0100, Reinhard Tartler encoded:
>> >> On Thu, Feb 03, 2011 at 20:13:09 (CET), Diego Elio Petten? wrote:
>> >> 
>> >> > Il giorno gio, 03/02/2011 alle 16.35 +0100, Anton Khirnov ha scritto:
>> >> >> 
>> >> >> it's been suggested on IRC that we've accumulated enough new APIs and
>> >> >> the associated cruft so the time to bump major for lavf and lavc is
>> >> >> nigh. We should definitely do that before 0.7.
>> >> >
>> >> > I'd suggest doing this only if we can also ensure that no ff_* symbols
>> >> > are left as interlib dependencies.
>> >> 
>> >> This sounds pretty reasonable to me.
>> >> 
>> >> As already mentioned, this is an excellent occasion for revisiting
>> >> Stefanos error code concern in avutil and the (potential?) avcore/avutil
>> >> merge, which both obviously are better done before bumping.
>> >
>> > BTW what's the best timing for doing the changes?
>> >
>> > I believed that deep changes are better done just *after* release,
>> > indeed I can imagine that many users just upgrade for the release:
>> >
>> > - Hey, these guys finally released a new FFmpeg, let's try it!
>> >
>> > - Ouch, it breaks a lot of stuff, better to keep the ol' good FFmpeg,
>> >   I don't have time to fix it now.
>> 
>> On the other hand, new releases are expected to bring new stuff.
>> 
>> > vs.
>> > - Well, they deprecated a lot of stuff and I'll have to cleanup later,
>> >   but it is already compiling *now*!
>> 
>> We could at least remove whatever was deprecated at the time of the
>> last release.  That's a long list already.
>> 
> I think $next release should throw out the stuff that's already
> deprecated, deprecate whatever we want to remove soon and isn't
> deprecated yet. (and if possible, it should contain -mt :) )
>
> The release notes should also state that after a short transitional
> period (0.5-0.75 y?) we'll do another bump and another release.

What problem would that solve?

We should bump major only when it makes sense and is necessary.

-- 
Gruesse/greetings,
Reinhard Tartler, KeyID 945348A4