[gst-devel] Re: Native Win32 build of GStreamer 0.8.1

[Benjamin Otte]

On Thu, 6 May 2004, Thomas Vander Stichele wrote:

> > I am curious, as we dont hear from you guys. Any news about the patch ?
> > Can you accept it as it is, or does it need some more changes ?
>
> I see some discussion going in between robux and ds on IRC.  What's the
> bug number so I can check ?
>
http://bugzilla.gnome.org/show_bug.cgi?id=141317

> > Please note that the complete team is waiting for some 'sign of life'
> > from you now, we wont progress on porting the plugins and making apps
> > before the patch was accepted.
>
> Well, to play devil's advocate, if the patch doesn't receive further
> communication either way (ie, with no unresolved "problems" on our
> side), *and* it doesn't mess up the unix-side of things, I'll commit it
> :)
>
The basic problem where there is no agreement is about what build
environment to use.
GStreamer requires a lot of files to be created during the built which
requires quite some tools being setup right. These tools include
among others autoconf (for gstconfig.h), bison, flex (the parser) and perl
(glib-mkenums). A typical Windows developer setup for Visual Studio has
none of those.
So the easy approach is to generate those files on a Linux box and include
them verbatim in the Windows build. This requires however that those files
get regenerated properly whenever something changes (which is the reason
why they are autogenerated in the first place).

The question is now: Should we require Windows developers to get a
complete build stack going with these tools or should we put workarounds
in our cvs to ease Windows developing?

Both David and me think we should make Windows developers create a
complete build stack while the Windows devs don't want that.
Which I can understand, because it's freakin hard to get that working in
Windows.

Benjamin