(toplevel) Don't look for unprefixed tools unless they're appropriate

(toplevel) Don't look for unprefixed tools unless they're appropriate

[Nathanael Nerode]

Tested on i686-pc-linux-gnu with a Canadian cross.  Now if the 
prefixed tools can't be found (and the unprefixed tools are incorrect), 
the tool values default to blank, which will cause entirely different 
errors.  (Heh.)

I'd rather have them default "correctly", but that's somewhat less
trivial, and it's really only useful for people who 'configure',
relocate, and then 'make', which doesn't work anyway.

	* configure.in: Don't default to unprefixed tools unless
	the native tools will work.
	* configure: Regenerate.

Index: configure.in
===================================================================
RCS file: /cvs/gcc/gcc/configure.in,v
retrieving revision 1.200
diff -u -r1.200 configure.in
--- configure.in	28 Dec 2002 06:57:48 -0000	1.200
+++ configure.in	28 Dec 2002 07:12:02 -0000
@@ -2008,15 +2008,15 @@
 AC_SUBST(config_shell)
 
 # Host tools.
-AC_CHECK_TOOL(AR, ar)
-AC_CHECK_TOOL(AS, as)
-AC_CHECK_TOOL(DLLTOOL, dlltool)
-AC_CHECK_TOOL(LD, ld)
-AC_CHECK_TOOL(NM, nm)
-AC_CHECK_TOOL(RANLIB, ranlib, :)
-AC_CHECK_TOOL(WINDRES, windres)
-AC_CHECK_TOOL(OBJCOPY, objcopy)
-AC_CHECK_TOOL(OBJDUMP, objdump)
+NCN_STRICT_CHECK_TOOL(AR, ar)
+NCN_STRICT_CHECK_TOOL(AS, as)
+NCN_STRICT_CHECK_TOOL(DLLTOOL, dlltool)
+NCN_STRICT_CHECK_TOOL(LD, ld)
+NCN_STRICT_CHECK_TOOL(NM, nm)
+NCN_STRICT_CHECK_TOOL(RANLIB, ranlib, :)
+NCN_STRICT_CHECK_TOOL(WINDRES, windres)
+NCN_STRICT_CHECK_TOOL(OBJCOPY, objcopy)
+NCN_STRICT_CHECK_TOOL(OBJDUMP, objdump)
 AC_SUBST(CC)
 AC_SUBST(CXX)
 AC_SUBST(CFLAGS)
@@ -2026,13 +2026,13 @@
 AC_SUBST(DEFAULT_M4)
 
 # Target tools.
-NCN_CHECK_TARGET_TOOL(AR_FOR_TARGET, ar)
-NCN_CHECK_TARGET_TOOL(AS_FOR_TARGET, as)
-NCN_CHECK_TARGET_TOOL(DLLTOOL_FOR_TARGET, dlltool)
-NCN_CHECK_TARGET_TOOL(LD_FOR_TARGET, ld)
-NCN_CHECK_TARGET_TOOL(NM_FOR_TARGET, nm)
-NCN_CHECK_TARGET_TOOL(RANLIB_FOR_TARGET, ranlib, :)
-NCN_CHECK_TARGET_TOOL(WINDRES_FOR_TARGET, windres)
+NCN_STRICT_CHECK_TARGET_TOOL(AR_FOR_TARGET, ar)
+NCN_STRICT_CHECK_TARGET_TOOL(AS_FOR_TARGET, as)
+NCN_STRICT_CHECK_TARGET_TOOL(DLLTOOL_FOR_TARGET, dlltool)
+NCN_STRICT_CHECK_TARGET_TOOL(LD_FOR_TARGET, ld)
+NCN_STRICT_CHECK_TARGET_TOOL(NM_FOR_TARGET, nm)
+NCN_STRICT_CHECK_TARGET_TOOL(RANLIB_FOR_TARGET, ranlib, :)
+NCN_STRICT_CHECK_TARGET_TOOL(WINDRES_FOR_TARGET, windres)
 
 AC_SUBST(GCC_FOR_TARGET)
 AC_SUBST(FLAGS_FOR_TARGET)

Re: (toplevel) Don't look for unprefixed tools unless they're appropriate

[Alexandre Oliva]

On Dec 28, 2002, Nathanael Nerode <neroden@twcny.rr.com> wrote:

> Tested on i686-pc-linux-gnu with a Canadian cross.  Now if the 
> prefixed tools can't be found (and the unprefixed tools are incorrect), 
> the tool values default to blank, which will cause entirely different 
> errors.  (Heh.)

Hmm...  I don't quite like a blank default.  I'd much rather default
to <host/target>-<tool>, since this at least gives the user the
opportunity to adjust PATH after configure, and gives more of an idea
of what's going on.

> 	* configure.in: Don't default to unprefixed tools unless
> 	the native tools will work.
> 	* configure: Regenerate.

Nevertheless, this patch is correct, and approved.  The change in the
default should be addressed as a separate patch.

Thanks,

-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer

Re: (toplevel) Don't look for unprefixed tools unless they're appropriate

[Doug Evans]

Alexandre Oliva writes:
 > On Dec 28, 2002, Nathanael Nerode <neroden@twcny.rr.com> wrote:
 > 
 > > Tested on i686-pc-linux-gnu with a Canadian cross.  Now if the 
 > > prefixed tools can't be found (and the unprefixed tools are incorrect), 
 > > the tool values default to blank, which will cause entirely different 
 > > errors.  (Heh.)
 > 
 > Hmm...  I don't quite like a blank default.  I'd much rather default
 > to <host/target>-<tool>, since this at least gives the user the
 > opportunity to adjust PATH after configure, and gives more of an idea
 > of what's going on.

Agreed!

(toplevel) Don't look for unprefixed tools unless they're appropriate

[Doug Evans]

Nathanael Nerode writes:
 > I'd rather have them default "correctly", but that's somewhat less
 > trivial, and it's really only useful for people who 'configure',
 > relocate, and then 'make', which doesn't work anyway.

What do you mean by "relocate"?

Re: (toplevel) Don't look for unprefixed tools unless they're appropriate

[Alexandre Oliva]

[-- Attachment #1: Type: text/plain, Size: 1322 bytes --]

On Dec 28, 2002, Doug Evans <dje@transmeta.com> wrote:

> Alexandre Oliva writes:
>> On Dec 28, 2002, Nathanael Nerode <neroden@twcny.rr.com> wrote:
>> 
>> > Tested on i686-pc-linux-gnu with a Canadian cross.  Now if the 
>> > prefixed tools can't be found (and the unprefixed tools are incorrect), 
>> > the tool values default to blank, which will cause entirely different 
>> > errors.  (Heh.)
>> 
>> Hmm...  I don't quite like a blank default.  I'd much rather default
>> to <host/target>-<tool>, since this at least gives the user the
>> opportunity to adjust PATH after configure, and gives more of an idea
>> of what's going on.

> Agreed!

Ok, this is what I'm checking in, after verifying that it does the
right thing on native, cross, host-x-host and host-x-build set ups,
namely, if host/target == build, default to the program name without
prefix, otherwise prefix it with the host/target tool prefix.

I predict we're going to have a *lot* of trouble when we switch to
autoconf 2.5x in the top level.  The way build/host/target_alias is
defined has changed, and we're going to have to duplicate the logic
autoconf uses for defaulting system names to pass the right arguments
down.  Eeek :-(  I knew that right when the changes were introduced in
autoconf, but I couldn't stop them, and then I had forgotten.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: toplev-conf.patch --]
[-- Type: text/x-patch, Size: 2827 bytes --]

Index: ChangeLog
from  Alexandre Oliva  <aoliva@redhat.com>
	* configure.in (host_configargs): Replace reference to
	no-longer-defined buildopts with --build=${build_alias}.
	* configure: Rebuilt.

Index: configure.in
===================================================================
RCS file: /cvs/gcc/gcc/configure.in,v
retrieving revision 1.201
diff -u -p -r1.201 configure.in
--- configure.in 28 Dec 2002 09:12:19 -0000 1.201
+++ configure.in 28 Dec 2002 17:28:42 -0000
@@ -1752,7 +1752,7 @@ case "${cache_file}" in
   cache_file_option="--cache-file=../${cache_file}" ;;
 esac
 
-host_configargs="${cache_file_option} ${buildopt} --host=${host_alias} --target=${target_alias} ${extra_host_args} ${baseargs}"
+host_configargs="${cache_file_option} --build=${build_alias} --host=${host_alias} --target=${target_alias} ${extra_host_args} ${baseargs}"
 
 target_configargs=${baseargs}
 
Index: config/ChangeLog
from  Alexandre Oliva  <aoliva@redhat.com>

	* acx.m4: Name cache variables properly.
	(NCN_STRICT_CHECK_TOOL): If program is not found and 
	value-if-not-found is empty, use ${ncn_tool_prefix}$2 or $2,
	depending on whether build != host or not.
	(NCN_STRICT_CHECK_TARGET_TOOL): Ditto, with the target prefix.

Index: config/acx.m4
===================================================================
RCS file: /cvs/gcc/gcc/config/acx.m4,v
retrieving revision 1.1
diff -u -p -r1.1 acx.m4
--- config/acx.m4 28 Dec 2002 06:57:49 -0000 1.1
+++ config/acx.m4 28 Dec 2002 17:28:42 -0000
@@ -24,9 +24,9 @@ if test -n "$ncn_target_tool_prefix"; th
                 [${ncn_target_tool_prefix}$2], , [$4])
 fi
 if test -z "$ac_cv_prog_$1" ; then
-  ncn_ct_$1=$$1
-  AC_CHECK_PROG([ncn_ct_$1], [$2], [$2], [$3], [$4])
-  $1=$ncn_ct_$1
+  ncn_cv_$1=$$1
+  AC_CHECK_PROG([ncn_cv_$1], [$2], [$2], [$3], [$4])
+  $1=$ncn_cv_$1
 else
   $1="$ac_cv_prog_$1"
 fi
@@ -45,11 +45,11 @@ if test -n "$ncn_tool_prefix"; then
 fi
 if test -z "$ac_cv_prog_$1" ; then
   if test $build = $host ; then
-    ncn_ct_$1=$$1
-    AC_CHECK_PROG([ncn_ct_$1], [$2], [$2], [$3], [$4]) 
-    $1=$ncn_ct_$1
+    ncn_cv_$1=$$1
+    AC_CHECK_PROG([ncn_cv_$1], [$2], [$2], [ifelse([$3],[],[$2],[$3])], [$4]) 
+    $1=$ncn_cv_$1
   else
-    $1="$3"
+    $1="ifelse([$3],[],[${ncn_tool_prefix}$2],[$3])"
   fi
 else
   $1="$ac_cv_prog_$1"
@@ -69,15 +69,13 @@ if test -n "$ncn_target_tool_prefix"; th
 fi
 if test -z "$ac_cv_prog_$1" ; then
   if test $build = $target ; then
-    ncn_ct_$1=$$1
-    AC_CHECK_PROG([ncn_ct_$1], [$2], [$2], [$3], [$4]) 
-    $1=$ncn_ct_$1
+    ncn_cv_$1=$$1
+    AC_CHECK_PROG([ncn_cv_$1], [$2], [$2], [ifelse([$3],[],[$2],[$3])], [$4]) 
+    $1=$ncn_cv_$1
   else
-    $1="$3"
+    $1="ifelse([$3],[],[${ncn_target_tool_prefix}$2],[$3])"
   fi
 else
   $1="$ac_cv_prog_$1"
 fi
 ]) []dnl # NCN_STRICT_CHECK_TARGET_TOOL
-
-

[-- Attachment #3: Type: text/plain, Size: 289 bytes --]


-- 
Alexandre Oliva   Enjoy Guarana', see http://www.ic.unicamp.br/~oliva/
Red Hat GCC Developer                 aoliva@{redhat.com, gcc.gnu.org}
CS PhD student at IC-Unicamp        oliva@{lsd.ic.unicamp.br, gnu.org}
Free Software Evangelist                Professional serial bug killer