Re: cpplib and iconv.h (bootstrap failure)

Re: cpplib and iconv.h (bootstrap failure)

[Zack Weinberg]

Gerald Pfeifer <gerald@pfeifer.com> writes:

> Hi Zack,
>
> do you think you'll have time to look into this regression caused by your
> patch before the next snapshot (on Wednesday)?

Please try the appended patch.  You will need to run autoconf and
autoheader.

I think this would also be addressed by updating to the latest iconv.m4,
but it requires autoconf 2.5x.  I'm not sure if the autoconf transition
ever got to the point where we could cut over the gcc directory.

zw

        * aclocal.m4 (AM_ICONV): Add explicit check for iconv.h.
        * cpphash.h, java/lex.h: Check both HAVE_ICONV and HAVE_ICONV_H
        before including iconv.h.

===================================================================
Index: aclocal.m4
--- aclocal.m4	26 Nov 2003 19:17:49 -0000	1.82
+++ aclocal.m4	1 Dec 2003 19:09:12 -0000
@@ -607,6 +607,8 @@ AC_DEFUN([AM_ICONV],
     done
    ])
 
+  AC_CHECK_HEADERS([iconv.h])
+
   AC_CACHE_CHECK(for iconv, am_cv_func_iconv, [
     am_cv_func_iconv="no, consider installing GNU libiconv"
     am_cv_lib_iconv=no
===================================================================
Index: cpphash.h
--- cpphash.h	2 Oct 2003 07:23:26 -0000	1.202
+++ cpphash.h	1 Dec 2003 19:09:12 -0000
@@ -25,7 +25,7 @@ Foundation, 59 Temple Place - Suite 330,
 
 #include "hashtable.h"
 
-#ifdef HAVE_ICONV
+#if defined HAVE_ICONV_H && defined HAVE_ICONV
 #include <iconv.h>
 #else
 #define HAVE_ICONV 0
===================================================================
Index: java/lex.h
--- java/lex.h	4 May 2003 14:05:15 -0000	1.35
+++ java/lex.h	1 Dec 2003 19:09:12 -0000
@@ -35,7 +35,7 @@ extern FILE *finput;
 /* A Unicode character, as read from the input file  */
 typedef unsigned short unicode_t;
 
-#ifdef HAVE_ICONV
+#if defined HAVE_ICONV_H && defined HAVE_ICONV
 #include <iconv.h>
 #endif /* HAVE_ICONV */
 

Re: cpplib and iconv.h (bootstrap failure)

[Daniel Jacobowitz]

On Mon, Dec 01, 2003 at 11:11:41AM -0800, Zack Weinberg wrote:
> Gerald Pfeifer <gerald@pfeifer.com> writes:
> 
> > Hi Zack,
> >
> > do you think you'll have time to look into this regression caused by your
> > patch before the next snapshot (on Wednesday)?
> 
> Please try the appended patch.  You will need to run autoconf and
> autoheader.
> 
> I think this would also be addressed by updating to the latest iconv.m4,
> but it requires autoconf 2.5x.  I'm not sure if the autoconf transition
> ever got to the point where we could cut over the gcc directory.

It did.  It won't be painless but if you touch base with Nathanael
there should be very little left.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: cpplib and iconv.h (bootstrap failure)

[Zack Weinberg]

Daniel Jacobowitz <drow@mvista.com> writes:

> On Mon, Dec 01, 2003 at 11:11:41AM -0800, Zack Weinberg wrote:
>> Gerald Pfeifer <gerald@pfeifer.com> writes:
>> 
>> > Hi Zack,
>> >
>> > do you think you'll have time to look into this regression caused by your
>> > patch before the next snapshot (on Wednesday)?
>> 
>> Please try the appended patch.  You will need to run autoconf and
>> autoheader.
>> 
>> I think this would also be addressed by updating to the latest iconv.m4,
>> but it requires autoconf 2.5x.  I'm not sure if the autoconf transition
>> ever got to the point where we could cut over the gcc directory.
>
> It did.  It won't be painless but if you touch base with Nathanael
> there should be very little left.

Nathanael, thoughts?

zw

Re: cpplib and iconv.h (bootstrap failure)

[Gerald Pfeifer]

On Mon, 1 Dec 2003, Zack Weinberg wrote:
>         * aclocal.m4 (AM_ICONV): Add explicit check for iconv.h.
>         * cpphash.h, java/lex.h: Check both HAVE_ICONV and HAVE_ICONV_H
>         before including iconv.h.

I tested this (by bootstrapping) under two different configurations, and
it this seems to work.  Thanks!

Gerald
-- 
Gerald Pfeifer (Jerry)   gerald@pfeifer.com   http://www.pfeifer.com/gerald/