Running Xvfb during a package build -- good idea?

[Daniel P. Berrange]

On Wed, Oct 01, 2008 at 08:58:29AM +0100, Richard W.M. Jones wrote:
> On Tue, Sep 30, 2008 at 08:35:59PM +0100, Richard W.M. Jones wrote:
> > It's a bit worse than that as well because I've just discovered that
> > this package [openssl] needs Wine to build as well as to run the
> > tests :-(
> 
> Dan ^^ ?
> 
> I can see two possible fixes for this issue in general:
> 
> (a) Set BuildArch i386, so it only builds on i386.  This is nasty
> because the output package ends up as i386, which it isn't (rpm
> doesn't know the difference between build architecture and host
> architecture).
> 
> [What's the difference between BuildArch and ExclusiveArch?]
> 
> Or:
> 
> (b) Work with upstream to remove the need to run executables during
> the build.
> 
> This (b) seems like a rare case where this project will benefit
> cross-compilers in general, because with cross-compilation you should
> never need to run built executables during a build.  For example if
> the target architecture was some ARM embedded chipset, there might not
> even exist an emulator to run the executables.

What exactly is required to be run during build ?

As an example, GTK / GLib have a requirement to run glib-genmarshal 
during their build process. The upstream source explicitly deals with
this in a cross-compile environment by running the native version
to generate the data for the cross-compile.  As such the mingw-gtk
package BuildRequires  the same version of the native 'gtk' RPM.

I'd imagine a similar solution would be sensible for openssl, though
without knowing what its doing I can't say for sure.


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 :|