Gilles Chanteperdrix wrote: > Jan Kiszka wrote: > > Gilles Chanteperdrix wrote: > > > Jan Kiszka wrote: > > > > Rodrigo Rosenfeld Rosas wrote: > > > > > Hi Jan, > > > > > > > > > > When compiling svn xenomai I noticed the following error with > --enable-dox-doc > > > > > switch: > > > > > > > > > > CC [M] kernel/xenomai/skins/rtdm/drvlib.o > > > > > kernel/xenomai/skins/rtdm/drvlib.c:1513:1: error: unterminated > #ifdef > > > > > > > > > > > > > I just passed this stage with trunk revision 671 without problems > > > > (doxygen 1.4.4). Moreover, there is no #ifdef or related preprocessor > > > > stuff in the latest drvlib.c at line 1513. > > > > > > I have a few warnings with doxygen 1.4.2: > > > > > > ksrc/skins/rtdm/drvlib.c:151: Warning: Member rtdm_task_destroy found in > multiple @defgroup groups! The member will be put in group rtdmtask, and not > in group rtdmtask > > > > That's totally weird: "You are in rtdmtask, and not in rtdmtask!?" Even > > more confusing, there is only a single "@defgroup rtdmtask" in the code. > > > > I think I said this before: doxygen's error reporting sucks! I'm not > > excluding that something is broken here, but I don't know where to look. > > Could anyone of you try some older revisions of drvlib.c and > > rtdm.h/rtdm_driver.h to check where the regression(?) may come from? > > I have just checked, the error appears with revision 121, an old > one. EXPORT_SYMBOL may be the culprit, again... >
It seems so. Doxygen 1.4.6, e.g., is messing up the RTDM groups.
I just played with a few workarounds, this one here seems to have the
effect we need:
PREDEFINED = DOXYGEN_CPP \
"EXPORT_SYMBOL(symbol)=struct {}"
Alternatively, I could move EXPORT_SYMBOL in drvlib.c out of all doxygen
groups, giving up the locality of function definition and symbol export
again. :(
Jan
signature.asc
Description: OpenPGP digital signature
_______________________________________________ Xenomai-core mailing list [email protected] https://mail.gna.org/listinfo/xenomai-core
