Re: [MinGW] Set NATIVE_SYSTEM_HEADER_DIR relative to configured prefix

[Bob Rossi <bob_rossi at cox dot net>]

On Mon, Dec 18, 2006 at 05:10:26PM -0500, Bob Rossi wrote:
> On Mon, Dec 18, 2006 at 01:06:30PM -0800, Brian Dessent wrote:
> > Bob Rossi wrote:
> > 
> > > OK. Thanks Brian. I'll admit this is new to me, so I could be making
> > > some stupid mistakes here. Basically, I needed wchar_t support with
> > > mingw, and I was told that this works with the 4.x series. Which indeed
> > > it does! So, I went to this page,
> > >   http://gcc.gnu.org/releases.html
> > > and downloaded the most recent release. Is that not the proper thing to
> > > do?
> > 
> > That's fine if you just want a stable compiler.  But the question you
> > are asking seems to have morphed into "does support for a relocatable
> > compiler on MinGW currently work" and for any kind of "what is the state
> > of the code" sort of thing, you want to look at the current code (aka
> > HEAD aka 4.3) and not the code on a release branch, which tends to be
> > fairly static.  Because if you find that something doesn't work, then
> > you are left wondering if it is still broken in the current sources or
> > if it was just broken on that release branch.
> 
> Well, I'm sliding fast down a very large hill. I just compiled/installed
> libgmp and libmpfr because svn needs those installed. Then, when I
> bootstrap head, I the below error.
> 
> Is this well known? Here is the comment there,
> 
>           /* Should this directory start with the sysroot?  */
>           if (sysroot && p->add_sysroot)
>             str = concat (sysroot, p->fname, NULL);
>           else if (!p->add_sysroot && relocated)
>             {
>               /* If the compiler is relocated, and this is a configured 
>                  prefix relative path, then we use gcc_exec_prefix instead 
>                  of the configured prefix.  */
>               gcc_assert (strncmp (p->fname, cpp_PREFIX,
>                                      cpp_PREFIX_len) == 0);

This is the value of the parameters that make this invocation of gcc
raise an assertion.

p->fname=/mingw/include cpp_PREFIX=C:/mingw cpp_PREFIX_len=8

Is this failing because one of them expects /mingw and the other expects
C:/mingw?

Is there a function in gcc that I can use to make the paths similar
before the comparison?

Note to reader: I'm trying to patch this myself, even though I have no
idea if this is supposed to raise an assertion for other reasons. Who
would I contact to help me solve this problem? This is with svn trunk.

Thanks,
Bob Rossi