[FFmpeg-devel] [PATCH 3/4] fftools/opt_common: stop accessing a private field

Alexander Strasser eclipse7 at gmx.net
Tue Oct 15 22:05:54 EEST 2024


On 2024-10-15 10:37 +0200, epirat07 at gmail.com wrote:
>
>
> On 15 Oct 2024, at 10:30, Anton Khirnov wrote:
>
> > Quoting Alexander Strasser via ffmpeg-devel (2024-10-15 07:34:26)
> >> On 2024-10-14 22:35 +0200, Anton Khirnov wrote:
> >>> Quoting Alexander Strasser via ffmpeg-devel (2024-10-14 22:21:38)
> >>>>> It is documented in the header.
> >>>>
> >>>> I figured as much but couldn't find a hint in in avfilter.h
> >>>>
> >>>> You changed it in a previous patch of this series or am I reading it wrong?
> >>>
> >>> I don't follow, I changed what?
> >>
> >> No, I don't mean you did sth.
> >>
> >> I'm just too stupid to read where in avfilter.h it is documented
> >> to be private.
> >>
> >> So I'm asking what is written where that indicates this.
> >
> > Here:
> > http://git.videolan.org/?p=ffmpeg.git;a=blob;f=libavfilter/avfilter.h;h=b88b31a834094f15d3159b017438499f1809b82f;hb=HEAD#l247
> >
> > Of course experience shows that such warnings do not really work, as
> > users ignore or fail to notice them.
>
> Yes because approaches like „everything below this line“
> falls apart horribly in Doxygen as it is completely lost there.
>
> Same for IntelliSense-like things in editors.

Thanks Anton for pointing to the exact line.

Now it all makes sense.

I was looking at the header and the AVFilter struct multiple times.

First at the beginning of the struct, then searching and later out
of paranoia paging up and down.

The thing is, I didn't expect nor see that comment as it was just
somewhere in between and looking like documentation of some other
field.

Then I went to the HTML rendered by Doxygen and like epirat07 pointed
out, of course couldn't find any hint either...


> > Which is why we have been switching
> > to an approach that outright removes private information from public
> > headers, but that first requires to eliminate its use.

Sure, I agree to that direction.

Still that functionality is useful and while it can be improved I
would hope for a solution that doesn't drop the feature now and
hopes for it to comeback at some later point.


  Alexander


More information about the ffmpeg-devel mailing list