Re: binutils-2.19 -- source tarball may be incomplete or inconsistent

[Keith Marshall]

On Tuesday 23 December 2008 09:14:59 Nick Clifton wrote:
> > Strangely, although I
> > didn't preserve the generated `deffilep.h', and it now appears
> > nowhere in either the source or the build tree, its absence
> > doesn't seem to affect the success of the build; is it a quirk of
> > building for the `mingw32' target, that it isn't required, or is
> > it just superfluous?
>
> Weird.  It is included by ld/pe-dll.[ch] which are used by the
> i386-mingw32 target, so I would have expected your build to
> complain. Possibly there is a missing dependency in the Makefile so
> that pe-dll.o does not know that it needs the deffile.h header.

Ah, but the file in question is deffilep.h, not deffile.h; the latter 
*is* present in the tarball, while the former is not, but appears to 
be generated, along with deffilep.c, from deffilep.y.  Furthermore, 
if I grep the source tree recursively for references to each of 
deffilep.h and deffile.h, I see the reference you mention for the 
latter file, but none at all, (other than in ld/Makefile.in), for the 
former, which leads me to suspect that it may not be required.

> > I plan to repackage binutils-2.19, for redistribution from the
> > MinGW site on SourceForge, and I would like to save other users
> > from the problems I have experienced.  Presumably:
> >
> >   * rescheduling the timestamps for bfd/doc/{elf.texi,bfd.info}
> >   * adding ld/deffilep.c
> >
> > are appropriate.
>
> Agreed.

Ok, thanks.  I did that.  The repackaged tarball may be found here:
https://sourceforge.net/project/showfiles.php?group_id=2435&package_id=11290

or downloaded directly from:
http://downloads.sourceforge.net/mingw/binutils-2.19-src.tar.gz

> >  Should I also add ld/deffilep.h?
>
> Yes I think so.

I included it anyway.  Even if it isn't strictly necessary, the 
additional overhead is small.

With these minimal packaging adjustments, the entire binutils package 
builds OOTB for MinGW, and so far has worked fine for me, (I use it 
cross-hosted on GNU/Linux); also, we have seen no negative reports 
from those using it natively.  AFAIK, this is the first major release 
to have achieved such OOTB status on MinGW; many thanks to all who 
have contributed to getting there.

Regards,
Keith.