On Thu, Sep 22, 2005 at  8:52:15AM +0200, Georg Baum wrote:
> So the segmentation fault happens when running uic. This should be
> reported to trolltech, because it seems to be a bug in uic. 

Just to follow up, I submitted a bug report on the Trolltech site,
and got this response:

> /konstruct/apps/office/lyx/work/lyx-1.3.6/src/frontends/qt2/ui'
> /usr/lib/qt3//bin/uic -tr qt_ -impl BiblioModuleBase.h
> BiblioModuleBase.ui -o
> +BiblioModuleBase.C   
> make[7]: *** [BiblioModuleBase.C] Segmentation fault
> Workaround: add -nounload to UICFLAGS

Thanks for your comments. It seems like the Lyx configure script is
trying to use qt 3 uic on qt 2 .ui files which is not fully supported.
Thus, if this is the case, we do not regard this as a bug.

Greetings,
Paal
----------------end of response----------------------

The configure log shows all library references to /usr/lib/qt3, as
expected, since QTDIR=/usr/lib/qt3, and shows:
  configure:13480: checking Qt version
  configure:13508: result: 3.3.4
which is correct for my installation.

But I did wonder about all those src/frontends/qt2/... pathnames in
LyX, which I noticed in Lyx 1.3.5.  I assumed the numbering scheme
was something internal to LyX, not to QT itself, but I can't really
tell from a naive examination of a few LyX qt2 source files.  Does
anyone here know if LyX 1.3.6 is using QT 2 .ui files, as Paal seems
to think?  If not, I should probably follow up on my bug report.

Since adding the -nounload flag to UICFLAGS makes LyX build
successfully, maybe it's not really a QT bug; maybe the LyX top-level
Makefiles should specify that flag.  I can't find any documentation
on either UICFLAGS or -nounload in any of the QT installation files,
or on the Trolltech site, so I can't speculate on reasons to not
always include that flag.  I noticed that flag was often used in
various postings to the Trolltech site, fwiw.  But there was never
any discussion of the flags themselves in those posts. 

Does anyone here know why LyX doesn't use -nounload by default? 

Cheers,

Jim

Reply via email to