This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[libstdc++] Do configury choies on target, not target_alias
- From: Phil Edwards <phil at jaj dot com>
- To: Ranjit Mathew <rmathew at hotmail dot com>
- Cc: libstdc++ at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
- Date: Thu, 3 Oct 2002 19:52:29 -0400
- Subject: [libstdc++] Do configury choies on target, not target_alias
- References: <an99bs$in1$1@main.gmane.org> <20020930094344.499a9984.bkoz@redhat.com> <anbgol$q7h$1@main.gmane.org> <20021002223814.B15804@disaster.jaj.com> <anglt3$d14$1@main.gmane.org> <20021003083408.C19006@disaster.jaj.com>
On Thu, Oct 03, 2002 at 08:34:08AM -0400, Phil Edwards wrote:
> On Thu, Oct 03, 2002 at 11:22:22AM +0530, Ranjit Mathew wrote:
> > OK, got it. I put in a few "echo's" in the configure script and
> > this is what I get (I explicitly specify "--target=mingw32"):
>
> I'm vaguely surprised that works. My impression was that --target should
> be specified as a full triplet, or at least a double.
>
>
> > target_alias=mingw32 (Both before and after the assigment above)
> > target=i386-pc-mingw32 (Orig. "mingw32", changed by config.sub to this)
>
> Right. 'target' is always changed to the official form; 'target_alias'
> is whatever the user typed.
>
>
> > IMO then the "case switches" later should either be based on "target"
>
> Probably, yes. I could have sworn we changed this already...
Nope, I only thought about changing it, then decided that target_alias
was safe. Guess I missed a few spots.
This only changes the switch stmts in configure.in to use the canonical form.
We continue to use target_alias in the staging include dir as the name of
the target-specific directory (because that /is/ safe), and for building
installation pathnames (because that's what GCC/toplevel currently does,
and we should match).
Also I updated the porting comments/docs.
The switch stmts only affect cross compilers. I tried building one using
the target in question ("--target=mingw32"), but configure died somewhere
in the binutils. So I'm not committing this until somebody else can
sanity-check it first, while I go hunt down binutils braindamage.
Phil
Index: configure.in
===================================================================
RCS file: /home/pme/Repositories/GCC/gcc/libstdc++-v3/configure.in,v
retrieving revision 1.105
diff -u -3 -r1.105 configure.in
--- configure.in 26 Sep 2002 05:25:05 -0000 1.105
+++ configure.in 3 Oct 2002 23:23:19 -0000
@@ -102,7 +102,7 @@
# GLIBCPP_CHECK_BUILTIN_MATH_SUPPORT
# GLIBCPP_CHECK_MATH_SUPPORT
- case "$target_alias" in
+ case "$target" in
*-linux*)
os_include_dir="os/gnu-linux"
AC_CHECK_HEADERS([nan.h ieeefp.h endian.h sys/isa_defs.h \
@@ -147,7 +147,7 @@
AC_DEFINE(HAVE_COPYSIGNF)
AC_DEFINE(HAVE_FREXPF)
AC_DEFINE(HAVE_HYPOT)
- case "$target_alias" in
+ case "$target" in
*-hpux10*)
AC_DEFINE(HAVE_FINITE)
AC_DEFINE(HAVE_FINITEF)
@@ -224,7 +224,7 @@
;;
esac
- case "$target_alias" in
+ case "$target" in
*-mingw32*)
;;
*-windiss*)