Discuss: Base packages for Win32 / Win64 / OS X cross-compilation

[Daniel P. Berrange]

On Wed, Feb 11, 2009 at 11:36:32PM +0100, Farkas Levente wrote:
> Richard W.M. Jones wrote:
> > I've noticed that the most recent mingw32-binutils package is in fact
> > using upstream binutils (it's shipped separately by mingw.org, but it
> > is identical to gnu.org binutils).  So the diagram should look like
> > this:
> > 
> > ------------------------------------------------------------------------
> >   mingw32-               mingw64-               darwinx-
> > ------------------------------------------------------------------------
> >   filesystem             filesystem             filesystem
> > 
> >   binutils  <--- ? --->  binutils               odcctools
> >     (from upstream)        (from upstream)        (from Apple)
> >  
> >   gcc       <--- ? --->  gcc                    gcc
> >     (from upstream)        (from upstream)        (from Apple)
> > 
> >   w32api                 headers                headers
> >     (from mingw)           (from mw64)            (from Apple)
> > 
> >   runtime                runtime                -
> >     (from mingw)           (from mw64)
> > ------------------------------------------------------------------------
> > 
> > Which raises also the possibility of combining mingw32-binutils and
> > mingw64-binutils together (as well as mingw32-gcc and mingw64-gcc as
> > mentioned in the previous email).
> 
> as always i prefer to keep as little package as possible even if the
> spec files are a bit different. even only one crosscompile-filesystem
> and one mingw-gcc etc.
> even on crosscompile-headers (even is we rename w32ai).
> may be such names:
> <prefix>-<platform>-<package>
> <prefix> ~ crosscompile or cross
> <platform> ~ win32, win64 and osx or darwinx
> <package> ~ zlib
> as currently mingw32 packages not really widespread we can easily rename
> everything than later. so it'd be useful to thing about it.
> let's discuss it now before f11!

No, this is far to late to consider doing this for Fedora 11. It would
create a very high risk of missing the F11 deadline and having nothing.
Focusing on Mingw32 for F11 as originally planned is best, and then
consider the more general problem for F12.

Regardless of whether we decide to rename now, or later, we'd need to
provide compatability with Provides/Obsoletes tags, so it doesn't
really matter when we do this - the pain will be the same.

Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|