On terça-feira, 30 de abril de 2013 18.30.08, Oswald Buddenhagen wrote: > On Tue, Apr 30, 2013 at 08:59:47AM -0700, Thiago Macieira wrote: > > On terça-feira, 30 de abril de 2013 17.24.29, Oswald Buddenhagen wrote: > > > > Can we do it somehow less magically? Isn't there a way to do it if it > > > > needs to be done, and not do it if it doesn't need to be done? > > > > > > no, because latest on the second in-source build it would not be > > > necessary any more, thus messing up the expectations of those who want a > > > rebuild. > > > > I'm seriously wondering why configure.exe needs to be rebuilt if none of > > the sources have changed. Can't we simply adapt the Makefile? > > that would mean shipping a complete build dir including object files. > also, the semantics get really funky for shadows builds (read-only > source dir).
No, it means adapting the Makefile so that no build is attempted if the final target (configure.exe) is newer than all the sources. The presence or absence of intermediate files and artifacts should not count. Is that possible? > > For the include dir, it's a little harder to determine, especially since > > this will also affect rules inside .prf files in mkspecs. Any chance we > > can simply try and run syncqt, but ignore errors? > > talking about "less magically", huh? A compromise. Run syncqt; ignore errors; fail if a key file like headers.pri isn't there. It will also help you with writing the rule for shadow builds: since there is no headers.pri in the shadow build yet, syncqt failing (because perl is missing) will be caught as an error. -- Thiago Macieira - thiago.macieira (AT) intel.com Software Architect - Intel Open Source Technology Center
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development
