[FFmpeg-devel] [PATCH 00/26] Major library version bump

[James Almer]

On 1/18/2023 4:28 PM, Anton Khirnov wrote:
> Quoting James Almer (2023-01-16 14:38:14)
>> It's been a while since the last bump, so it's time to do some cleaning and
>> remove deprecated APIs. This will also give us an "Open ABI season" in which we
>> can do breaking changes (like changing public struct offsets, public enum
>> values, adding fields to structs that have their size tied to the ABI, etc) for
>> a few weeks.
> 
> Last time this open season lasted something like half a year and only
> ended when I arbitrarily said it did.
> 
> So I'd suggest to decide right now how long will the instability period
> last (6 weeks should be enough for everybody) and write the end date at
> the top of doc/APIchanges.
> 
> Another thing I'm not entirely happy about is versioning during the bump
> and instability. While the remove-then-bump approach does make bisection
> easier, it also creates commits that lie about their ABI version.

Does it really matter? All the patches will be pushed at the same time, 
meaning one git fetch will give you a stable state pre bump and the next 
will be right after it.
I think it's a bit farfetched to expect someone to pick a random commit 
in the middle of the bump and try to use the resulting compiled 
libraries with some program that was linked to some earlier version 
libraries.

> 
> I wonder if we couldn't come up with a better soltion. One thing that
> comes to mind is setting the major version to 0 until the instability
> period ends.

This could have several undesired effects, mainly for users looking at 
that define and not really expecting such value (There are several 
projects supporting more than one ffmpeg release and "MAJOR <= xx" 
checks are commonplace).
Also, if we are going to code the instability period in some form into 
the codebase, might as well make it so it starts with the first removal 
commit, or immediately before it, so what you described above is no 
longer a concern.

> 
>> I'm also taking this opportunity to suggest a change in our deprecation period
>> policy. Until now it's been a generic two years period, with no concrete reason
>> for it other than giving library users "time" to migrate. What we have seen
>> however is that users migrate in two cases: As soon as things are deprecated
>> when they use git head to get rid of deprecation warnings, or when they have no
>> choice (aka, when they want to move their project to a new ffmpeg version that
>> no longer has the symbols they depended on).
>> In the latter case, any arbitrary amount of time will make no difference
>> whatsoever. Projects could right now still be using ffmpeg 4.3 (since that's
>> what Debian stable ships) and would not consider moving to 5.1 or any future
>> version for the foreseeable future. So the suggestion is to change to a release
>> based scheme, which will in some form be time based anyway. Namely, every three
>> releases we do a major bump, which will be a good year or so in real world
>> terms, in which all API deprecated during that period, as long as it's present
>> in a release, is removed. This would also go with the idea of a recurrent LTS
>> release, so if we do three releases per major version, it could be x.0 (initial
>> release) x.1 (LTS), and x.2 (last release made pre bump).
> 
> Sounds good to me.
> 
>> If we go the above route, we could also remove API like the old lavu FIFO stuff,
>> a deprecation that's slightly less than a year old but effectively present in
>> v5.1.
>> We'd also need to add all this in writing, because this kind of policy can't
>> just be "oh yeah, we do it that way" in random emails.
> 
> But folklore is the most time-tested method of transmitting information.
>