On Sat, 20 Sep 2025 14:12:34 +0200,
Rafael Sadowski <rafael@sizeofvoid.org> wrote:
> 
> On Sat Sep 20, 2025 at 12:55:44PM +0100, Stuart Henderson wrote:
> > IMO it is worth committing diffs to preemptively fix ports with
> > newer FFmpeg before the update, as long as they don't break with
> > the old version - I only skimmed the diff but it looked like
> > these are generally safe from a quick look.
> > 
> > (also if there is a choice between fixing by adding a patch, and
> > fixing by updating the relevant port, updating would generally be
> > preferred)
> > 
> 
> Yes, that worked really well last time. I would be in favour of
> committing these changes only after building with -current and
> then having a small diff to test it.
> 

Thanks for feedback. It was original design of patches but at some point I
realized that moving with one large patch is simpler way to manage it.

Neverless, I had split it into 3 parts:
 - ffmpeg8.diff contains an updated for ffmpeg only
 - ffmpeg8-drop-postprocess.diff contains drop of postprocess library where
   it is hardcoded and from WANTLIB because it was removed at ffmpeg8
 - ffmpeg8-patches.diff patches which allows me to build ports against both,
   ffmpeg6 (from ports) and ffmpeg8 (from a diff)

I've attached all of them as gzip.

The bulk with only ffmpeg8-patches.diff ended as:

Elapsed time=1 day 19:31:15
I=12146 B=0 Q=0 T=0 F=0 !=109
L=textproc/xmlwf www/tuba

where:
 - textproc/xmlwf fails due to undefined symbol: XML_SetAlloc...
 - www/tuba fails due to duplicating `BreakpointCondition' instance...

both aren't related to ffmpeg.

-- 
wbr, Kirill

• ffmpeg8.diff.gz (8K)
• ffmpeg8-drop-postprocess.diff.gz (3K)
• ffmpeg8-patches.diff.gz (47K)