[patch] configure --without-x should work

[patch] configure --without-x should work

[Nathanael Nerode]

I've been told that the insight people are the ones to ask about this
patch to top level configure.in...

This deals with a little problem I discovered after autoconfiscating.
Since in my brave new autoconfiscated universe, dependencies in the 
Makefile will be *real* dependencies, stuff breaks until I change this.

What this does is to add to the behavior when --without-x is specified.  
tix and itcl won't behave properly without tk, let alone X, and the
dependencies of gdb on tk & company (which are optional) won't be properly 
satisfied without tk being present.

--without-x is currently broken (doesn't build), so this is no loss.

I'd like review from someone who knows something about the whole with-x/
without-x thing.

Index: configure.in
===================================================================
RCS file: /cvs/gcc/gcc/configure.in,v
retrieving revision 1.164
diff -u -r1.164 configure.in
--- configure.in	1 Jul 2002 21:01:48 -0000	1.164
+++ configure.in	1 Jul 2002 21:06:01 -0000
@@ -175,7 +175,11 @@
 
 case ${with_x} in
   yes | "") ;; # the default value for this tree is that X11 is available
-  no)  skipdirs="${skipdirs} tk libgui" ;;
+  no)
+    skipdirs="${skipdirs} tk tix itcl libgui"
+    # We won't be able to build gdbtk without X.
+    enable_gdbtk=no 
+    ;;
   *)  echo "*** bad value \"${with_x}\" for -with-x flag; ignored" 1>&2 ;;
 esac
 

Re: [patch] configure --without-x should work

[Keith Seitz]

On Wed, 3 Jul 2002, Nathanael Nerode wrote:

> --without-x is currently broken (doesn't build), so this is no loss.

Agreed.

> I'd like review from someone who knows something about the whole with-x/
> without-x thing.

I think this is fine. It makes perfect sense. There is only one really,
really, really minor point worth mentioning, though: cygwin. On cygwin, it
is possible to have X but not want to build with it, i.e., use the default
cygwin builds, which link against Microsoft DLL import libraries.

However, I don't think that any of our configure machinery would work in
this case, anyway, so it's still no loss. :-)

Keith

Re: [patch] configure --without-x should work

[Nathanael Nerode]

On Wed, Jul 03, 2002 at 01:27:10PM -0700, Keith Seitz wrote:
> On Wed, 3 Jul 2002, Nathanael Nerode wrote:
> 
> > --without-x is currently broken (doesn't build), so this is no loss.
> 
> Agreed.
> 
> > I'd like review from someone who knows something about the whole with-x/
> > without-x thing.
> 
> I think this is fine. It makes perfect sense. There is only one really,
> really, really minor point worth mentioning, though: cygwin. On cygwin, it
> is possible to have X but not want to build with it, i.e., use the default
> cygwin builds, which link against Microsoft DLL import libraries.
> 
> However, I don't think that any of our configure machinery would work in
> this case, anyway, so it's still no loss. :-)

Well, I was told that if it *does* use our configure machinery, it
uses it with --with-x !  I'd love to understand how *that* works, but
I'd rather put it off until after autoconfiscation.

Soooo, can someone approve me? ;-)

--Nathanael

Re: configure --without-x should work

[DJ Delorie]

> Well, I was told that if it *does* use our configure machinery, it
> uses it with --with-x !  I'd love to understand how *that* works, but
> I'd rather put it off until after autoconfiscation.
> 
> Soooo, can someone approve me? ;-)

Keith, I'll approve it on the gcc list if you say it's OK...