On Wed, 30 Mar 2011, Måns Rullgård wrote:

> Martin Storsjö <mar...@martin.st> writes:
> 
> > On Wed, 30 Mar 2011, Måns Rullgård wrote:
> >
> >> Martin Storsjö <mar...@martin.st> writes:
> >> 
> >> > On Wed, 23 Mar 2011, Måns Rullgård wrote:
> >> >
> >> >> Martin Storsjö <mar...@martin.st> writes:
> >> >> 
> >> >> > On Wed, 23 Mar 2011, Måns Rullgård wrote:
> >> >> >
> >> >> >> Martin Storsjö <mar...@martin.st> writes:
> >> >> >> 
> >> >> >> > +        enable dos_paths
> >> >> >> > +        ASFLAGS=$(echo $(filter_out --include* $ASFLAGS))
> >> >> >> 
> >> >> >> filter_asflags="filter_out --include*"
> >> >> >
> >> >> > Ah yes - I tried that initially, but this is past the section where 
> >> >> > cflags 
> >> >> > are added to ASFLAGS ("add_asflags $extra_cflags", quite soon after 
> >> >> > the 
> >> >> > compiler detection section) - setting filter_asflags here doesn't 
> >> >> > have any 
> >> >> > effect.
> >> >> >
> >> >> > Do you prefer moving part this up a bit, as a separate block, next to 
> >> >> > the 
> >> >> > compiler section? Something like this:
> >> >> >
> >> >> > if [ "$target_os = "symbian" ]; then
> >> >> >     filter_asflags="filter_out --include*"
> >> >> > endif
> >> >> 
> >> >> No, that's worse.  Bother me about it later today, I'll try to think of
> >> >> a nice solution.
> >> >
> >> > Ping - any better suggestion?
> >> 
> >> What does the --include do, and can we add it ourselves somehow?  That
> >> would solve the problem and make life easier for the user.
> >
> > The --include is for including a toolchain (gcc or rvct) specific header 
> > that all the platform headers require to be present before they are 
> > included.
> >
> > If we'd add it ourselves, it would be something like this for the gcce 
> > case:
> >
> > add_cflags --include=$EPOCROOT/epoc32/include/gcce/gcce.h
> >
> > It's kinda ugly to have to rely on an external env variable for the SDK 
> > root, although $EPOCROOT is the standard env var name for this in symbian 
> > stuff.
> 
> Can we use the existing --sysinclude flag for this?

Yes, that seems to work.

Tests like _FILE_OFFSET_BITS=64 that are done before the target-os switch 
(where the include flag is added to cflags) do fail even though they would 
have succeeded otherwise, but that's not really an issue, I'm not sure the 
os itself supports such files anyway.

// Martin
_______________________________________________
libav-devel mailing list
libav-devel@libav.org
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to