[FFmpeg-devel] [PATCH] make img_convert symbol conditional on lavc version, not libswscale

[Michael Niedermayer]

On Sat, Jul 05, 2008 at 03:32:49PM +0200, Diego Biurrun wrote:
> On Mon, Jun 30, 2008 at 11:34:41PM +0200, Michael Niedermayer wrote:
> > On Mon, Jun 30, 2008 at 10:55:53PM +0200, Reinhard Tartler wrote:
> > > Michael Niedermayer <michaelni at gmx.at> writes:
> > > > The question is, do we have someone who is willing and able to maintain
> > > > it? One who will fix bugs reported by users in case there are any ...
> > > > So far people just said they want the patch, also at least one said
> > > > that he wouldnt be able to maintain it. I do not remember anyone
> > > > volunteering to maintain this patch ...
> > > 
> > > What qualification would that person need?
> > 
> > able to code in C, review C code use a debugger, compiler and editor and be
> > smart enough to debug possibly non trivial bugs in a C program.
> > With some luck the patch is fine but if not there could be various obscure
> > issues
> > IIRC some issue was actually already found and fixed in some variant used
> > by some distro in this thread ...
> 
> My patch is better than just adding '#if 0' in some places, which is
> what some distros did.
> 
> > > AFAIUI Diego seems to be willing to support it as he proposed it in the
> > > beginning of this thread. I'm willing to assist him with that from the
> > > distro POV.
> > > 
> > > >> Distros are carrying this patch for years now without serious
> > > >> concerns, after all, so it cannot be that broken as one might suspect
> > > >> in the first place.
> > > >
> > > > Distros are carrying alot for years ...
> > > >
> > > > My general rule of thumb is nowadays, if something doesnt work and its not
> > > > obviously my own fault, it tends to be the vendors fault more often than
> > > > upstreams ...
> > > 
> > > That's why this patch was forwarded to upstream in the first place: In
> > > the hope to get commented properly. Up to know the only answer I got
> > > was "naah, go away, we don't know what effects this might have and don't
> > > care enough to look at it". :/ [1]
> > 
> > If we have someone willing to maintain then his first job is to properly
> > review the patch take a quick look at old and new stuff and declare if its
> > ok or not (and if not then why not) ...
> 
> Somehow I fail to see what exactly the problem with my patch is.  

Ive said it a few times already, I like to spend my time moving ffmpeg
forward, these patches do not move us forward. And neither do they
move applications using ffmpeg forward.
They rather make a case that should not be used more useable.
If someone wants this, he can review and maintain these #if 0 patches and
apply them, i will NOT spend a day reviewing the scalers and the code
using them to check if these hacks might work.
The reason is simply that i could make ffmpeg with the regresson tests fully
functional with the new scaler in the same time. And then drop the old
one.


> Right
> now we *do* have a problem.  The problem is in our code, not in some
> distro tree.

And there are many solutions:
* remove the old scaler with a version bump
* mark --enable-swscale as experimental and not for production use
* bump the version when --enable-swscale is used.

If we really want to allow both scalers to be enabled we at least need
someone reviewing the scalers and code using them to ensure they dont
conflict. Noone seems to volunteer so what is it that you want?

That i should NOT fix all remaioning issues in the new scaler but rather
figure out a way to make the old scaler (slow) and the new (half working)
scaler supported at the same time?


>  We have a buggy mix of conditionals.  Some parts of the
> code depend on lavc version, others on libswscale.  This cannot be
> right.
> I have a feeling that you are going off on a tangent.  What exactly is
> the problem with my patch?


[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Many that live deserve death. And some that die deserve life. Can you give
it to them? Then do not be too eager to deal out death in judgement. For
even the very wise cannot see all ends. -- Gandalf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080705/73bdbe7c/attachment.pgp>