Non-functional compiler under SGI Irix

[Philippe Schaffnit]

Thanks a lot for your help!

It's filed as [Bug bootstrap/30272] (I couldn't find 'fixinclude').

Philippe

Steve Kargl wrote:
> 
> On Wed, Dec 20, 2006 at 08:14:45PM +0100, Daniel Franke wrote:
> > On Wednesday 20 December 2006 16:30, Steve Kargl wrote:
> > > On Wed, Dec 20, 2006 at 08:55:48AM +0100, Philippe Schaffnit wrote:
> > > > Does it say anything to anyone?
> > >
> > > Yes.  It says that Irix's header files are screwed up.
> > >
> > > > /USER/philippe/Irix/Gcc_Sources/libgfortran/fmain.c:2:
> > > > /usr/include/complex.h:178: error: conflicting types for 'cabsl'
> > > > /USER/philippe/Irix/Compilation/Temp/./gcc/include/math.h:537: error:
> > > > previous declaration of 'cabsl' was here
> > >
> > > Why is cabsl in math.h?  I believe it should be in complex.h.
> >
> > >From fixincludes/inclhack.def:1020-1028:
> >
> > /*
> >  *  Remove `extern double cabs' declarations from math.h.
> >  *  This conflicts with C99.  Discovered on AIX.
> >  *  IRIX 5 and IRIX 6 before 6.5.18 (where C99 support was introduced)
> >  *  declares cabs() to take a struct __cabs_s argument.
> >  *  SunOS4 has its cabs() declaration followed by a comment which
> >  *  terminates on the following line.
> >  *  Darwin hides its broken cabs in architecture-specific subdirs.
> >  */
> >
> > There's nothing about `cabsl` in that file ...
> >
> 
> Open a fixinclude bug report.  cabsl is the "long complex" version
> in the cabsf, cabs, and cabsl family of functions.  Seems like it
> would be a one liner type patch.
> 
> --
> Steve