[FFmpeg-devel] [PATCH] lavc: remove the QOA decoder

[Tomas Härdin]

fre 2023-12-22 klockan 10:09 -0300 skrev James Almer:
> On 12/21/2023 5:05 PM, Paul B Mahol wrote:
> > On Thu, Dec 21, 2023 at 8:43 PM Tomas Härdin <git at haerdin.se>
> > wrote:
> > 
> > > ons 2023-12-20 klockan 20:11 +0100 skrev Michael Niedermayer:
> > > > On Wed, Dec 20, 2023 at 05:57:40PM +0100, Tomas Härdin wrote:
> > > > > tis 2023-12-19 klockan 15:02 +0100 skrev Nicolas George:
> > > > [...]
> > > > > [...] , but every line of code
> > > > > carries with it a non-zero maintenance burden
> > > > 
> > > > Assuming you mean with "non-zero" a "larger than zero"
> > > > maintenance
> > > > burden
> > > > 
> > > > then we can proof this to be false
> > > 
> > > Doubt
> > > 
> > > > What iam trying to say is, the maintaince burden resulting from
> > > > a
> > > > change
> > > > is complex
> > > 
> > > Indeed
> > > 
> > > > In this specific case here we have a patch proposing the
> > > > removial of
> > > > a decoder
> > > > missing a test.
> > > > Its easy to say the burden is less when the decoder is removed
> > > > But its author recently left the project too
> > > 
> > > This is one problem. But the careless attitude to shoving more
> > > features
> > > into the codebase is far more serious. Every line of code is a
> > > CVE
> > > waiting to happen. Apparently this is a difficult thing to grasp
> > > for
> > > some contributors. It's an attitude I expect only from junior
> > > developers.
> > > 
> > > Ensuring C code is correct and safe is *hard*. I have spent time
> > > formally verifying embedded C code for spaceflight. The lessons
> > > learned
> > > from this has made me supremely suspicious of cowboy coding.
> > > 
> > > I have raised this issue multiple times in this project to no
> > > avail. I
> > > do not expect things to change.
> > > 
> > 
> > Say what serious feature you contributed ? - Nothing.
> 
> First of all, this was completely uncalled for. Second, mxf support
> in 
> lavf is pretty much thanks to him.

I will add that Baptiste is mostly to thank for it. But I'd prefer if
it were replaced by bmxlib. This would allow concentrating MXF dev
effort on a single project. But already at the concept stage there has
been opposition to this.

/Tomas