On Fri, Apr 26, 2013 at 02:30:10PM -0700, Thiago Macieira wrote:
> With that, Ossi's patch is one step in the right direction and one step in
> the
> wrong direction. We should not re-generate the headers if they are present,
> but we should still do it if they aren't -- and that includes shadow builds.
>
my current solution is entirely predictable: git builds always run
syncqt, while other builds never do. this makes a very clear statement:
if you want to modify qt (or at least its apis), use git.
running it based on existence checks is a mess. what about interrupted
runs? what about headers that were added later?
shadow builds are a different matter ... the current solution of
copying/symlinking stuff to the builddir is pretty sick. the idea was
making the build dir a complete installation dir, but this doesn't
*really* work anyway - you still need the source tree around. that means
that we can actually make full use of the source tree - qmake is now
clever enough to do that (c.f., mkspecs/), and dealing properly with
includes in the source tree is also doable (duh). having said that, the
easiest solution is changing the condition to "git or shadow build" to
make it work and still be entirely predictable.
_______________________________________________
Development mailing list
Development@qt-project.org
http://lists.qt-project.org/mailman/listinfo/development
