[Psi-Devel] psi 0.14 build error

Justin Karneges justin-psi2 at affinix.com
Thu Dec 17 23:45:58 PST 2009


On Thursday 17 December 2009 22:51:02 Thilo Cestonaro wrote:
> Am 18.12.2009 um 01:04 schrieb Justin Karneges:
> > Hmm, there may be another problem here, in that qmake -prl is executed
> > before conf.pri is written.  This means that on the first run the config
> > is totally wrong, and on subsequent runs the previous config is used.
> >
> > I'll think about how to rework this so conf.pri exists when qmake -prl is
> > invoked.  I don't know if this would solve the bug being discussed
> > though.
>
> on Windows the first configure (qmake) run produces an segfault too.
> But the second works fine.
>
> So I guess it is the same problem.

I've attempted a fix in git.  I copied the conf.pri generation code from qconf 
itself into recursiveprl.qcm so that it can create the file early.  The 
configure script still redundantly writes conf.pri again at the end of all 
processing, even though it will already exist with the same content.  
Definitely a hack, but it should work.

I don't know yet what the "right" fix is.  The simplest answer would be to 
have a special flag in qconf that tells it to run qmake -prl after conf.pri 
is generated but before qmake (without -prl) is run.  That would at least 
give the configure script the right behavior without any hacks, but I dislike 
the idea of having such a "special" flag.  Maybe a better answer would be to 
make qconf post-processing generally customizable.

-Justin


More information about the Psi-Devel mailing list