[vlc-devel] commit: Fix live555 locale patch once again. (Pavlov Konstantin )

[Rémi Denis-Courmont]

On Friday 27 March 2009 02:52:28 Ross Finlayson wrote:
> >No thanks. I'd rather keep the patches in the VLC contribs then.
>
> But Rémi, if the proposed changes are of general
> applicability, rather than just being
> VLC-specific (and it they're VLC-specific, then
> they should be made to
> "modules_demux_live555.cpp" instead, because
> that's the only LIVE555-related code that's
> supposed to be VLC-specific), then I don't
> understand why you would not want us to make the
> changes to the released version of LIVE555, so
> that you won't have to worry about patching it
> again.

If it were _easy_ to patch live555, then we could do that. But for lack of a 
proper publicly available Version Control System, it is not easy.

In addition to that, we do have non-general patches. The locale() patches are 
POSIX.2008-compliant, but we all know they break lots of platforms, especially 
Windows. The VideoLAN project has no interest in addressing that portability 
problem in a way that is likely to be acceptable to upstream (i.e. You).

BSD projects have ports with BSD-specific patches. Linux distributions have 
distribution or Linux-specific patches in their packages. And you can bet 
proprietary software vendors have proprietary patches on top of the BSD/MIT-
licensed components that they are sourcing from the open-source community.

Certainly, the VideoLAN project could do a better at "pushing patches" 
upstream, but this cannot be a one-sided effort. And in any case, those 
patches will continue to _first_ go to our contrib, before they are forwarded 
upstream.

> But if people aren't willing to even discuss with
> us why they think they need these changes, then
> they're just making things more difficult for
> themselves - especially since, as far as I can
> tell, the proposed 'Locale' changes are bogus,
> because they're going to break on many OSs for
> which LC_NUMERIC_MASK is not even defined.

The explanation was provided when the patch was initially written..

> If the VLC maintainers want to make more work for
> themselves by forcing themselves to update these
> patches every time we change our code, then
> that's fine with me, but I'd prefer to help save
> some of their time, as well as get the benefit of
> possible general bug fixes (because there are
> many other applications - beyond VLC - that use
> our code).

FYI, my motivation for pushing patches is that we do not control the "contrib" 
on Linux ports of VLC. That is way more problematic for us than some extra 
work when we update the contrib.

-- 
Rémi Denis-Courmont