Re: 3.1 branch ada bootstrap fails

Re: 3.1 branch ada bootstrap fails

[Simon Wright]

(Sorry this isn't threaded, I've only just joined)

A "drill" solution might be to alter the GNAT section of the
installation instructions to recommend configuring with the old GNAT
compiler:

  CC=gnatgcc; export CC
  {srcdir}/configure --enable-languages=c,ada

Re: 3.1 branch ada bootstrap fails

[Mark Mitchell]

--On Tuesday, May 07, 2002 08:49:13 PM -0300 Alexandre Oliva 
<aoliva@redhat.com> wrote:

> On May  7, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
>
>> Could someone else check and install this patch?
>
> I'm on it.

Thanks.

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> On May  7, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
> 
> > autoconf: Undefined macros:
> > configure.in:457:gcc_AC_PROG_LN_S
> 
> I didn't get this.  Which variant of autoconf 2.13 did you use?

Vanilla.  The problem actually is m4-1.4o.  m4-1.4 is ok.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: 3.1 branch ada bootstrap fails

[Alexandre Oliva]

On May  7, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:

> autoconf: Undefined macros:
> configure.in:457:gcc_AC_PROG_LN_S

I didn't get this.  Which variant of autoconf 2.13 did you use?

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

Re: 3.1 branch ada bootstrap fails

[Alexandre Oliva]

On May  7, 2002, Alexandre Oliva <aoliva@redhat.com> wrote:

> On May  7, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
>> Could someone else check and install this patch?

> I'm on it.

Ok, I've just verified it fixed the problem I had with yesterday's
snapshot, so I'm checking it in.

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

Re: 3.1 branch ada bootstrap fails

[Alexandre Oliva]

On May  7, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:

> Could someone else check and install this patch?

I'm on it.

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

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> > <http://gcc.gnu.org/ml/gcc/2002-05/msg00286.html>.
> 
> That patch is fine; please apply it.

I extracted Geoffrey's patch and updated it so that it applies cleanly
on the branch.  However, there seems to be a problem regenerating
configure with autoconf 2.13.  I get the following warning:

autoconf: Undefined macros:
configure.in:457:gcc_AC_PROG_LN_S

I did a check comparing the current version of configure with a newly
generated version from the current version of configure.in.  There
were a couple of different lines that I didn't like.  This suggests
an autoconf problem (maybe a shell issue) under hpux.

Could someone else check and install this patch?

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

2002-05-07  Geoffrey Keating  <geoffk@redhat.com>

	* configure.in: Don't pass -Wno-long-long to a ADA compiler
	that doesn't support it.
	* configure: Regenerate.

Index: configure.in
===================================================================
RCS file: /cvs/gcc/gcc/gcc/configure.in,v
retrieving revision 1.576.4.13
diff -u -3 -p -r1.576.4.13 configure.in
--- configure.in	7 May 2002 00:15:06 -0000	1.576.4.13
+++ configure.in	7 May 2002 02:54:00 -0000
@@ -349,6 +349,9 @@ fi
 AC_SUBST(NO_MINUS_C_MINUS_O)
 AC_SUBST(OUTPUT_OPTION)
 
+# See if GNAT has been installed
+gcc_AC_PROG_GNAT
+
 AC_CACHE_CHECK(whether ${CC-cc} accepts -Wno-long-long,
 ac_cv_prog_cc_no_long_long,
 [save_CFLAGS="$CFLAGS"
@@ -356,8 +359,26 @@ CFLAGS="-Wno-long-long"
 AC_TRY_COMPILE(,,ac_cv_prog_cc_no_long_long=yes,
 	       ac_cv_prog_cc_no_long_long=no)
 CFLAGS="$save_CFLAGS"])
+
+if test x$have_gnat != xno ; then 
+AC_CACHE_CHECK(whether ${ADAC} accepts -Wno-long-long,
+ac_cv_prog_adac_no_long_long,
+[cat >conftest.adb <<EOF
+procedure conftest is begin null; end conftest;
+EOF
+if $ADAC -Wno-long-long -c conftest.adb 1>&5 2>&5 ; then
+  ac_cv_prog_adac_no_long_long=yes
+else
+  ac_cv_prog_adac_no_long_long=no
+fi
+rm -f conftest*])
+else
+  ac_cv_prog_adac_no_long_long=yes
+fi
+
 strict1_warn=
-if test $ac_cv_prog_cc_no_long_long = yes; then
+if test $ac_cv_prog_cc_no_long_long = yes && \
+    test $ac_cv_prog_adac_no_long_long = yes ; then
   strict1_warn="-pedantic -Wno-long-long"
 fi
 AC_SUBST(strict1_warn)
@@ -476,9 +497,6 @@ AC_CHECK_HEADER(pthread.h, [have_pthread
 gcc_AC_C_CHAR_BIT
 gcc_AC_C_COMPILE_ENDIAN
 gcc_AC_C_FLOAT_FORMAT
-
-# See if GNAT has been installed
-gcc_AC_PROG_GNAT
 
 # See if we have the mktemp command.
 AC_CHECK_PROG(have_mktemp_command, mktemp, yes, no)

Re: 3.1 branch ada bootstrap fails

[Mark Mitchell]

--On Monday, May 06, 2002 06:25:56 PM -0400 John David Anglin 
<dave@hiauly1.hia.nrc.ca> wrote:

>> >> 2) Port from trunk Geoffrey Keating's configure patch to check if
>> >>    -Wno-long-long is supported by the ada driver.
>> >
>> > I'm in favor of (2), for what it's worth; we know it works on the
>> > trunk.
>>
>> Do you have a pointer to (2)?
>
> <http://gcc.gnu.org/ml/gcc/2002-05/msg00286.html>.

That patch is fine; please apply it.

Thanks,

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> >> 2) Port from trunk Geoffrey Keating's configure patch to check if
> >>    -Wno-long-long is supported by the ada driver.
> >
> > I'm in favor of (2), for what it's worth; we know it works on the
> > trunk.
> 
> Do you have a pointer to (2)?

<http://gcc.gnu.org/ml/gcc/2002-05/msg00286.html>.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: 3.1 branch ada bootstrap fails

[Mark Mitchell]

--On Monday, May 06, 2002 01:28:10 PM -0700 Zack Weinberg 
<zack@codesourcery.com> wrote:

> On Mon, May 06, 2002 at 11:12:35AM -0400, John David Anglin wrote:
>> > I am not sure whether ALL_CFLAGS wouldn't have other junk, such as
>> > unwanted -D switches.  However, if it doesn't, it is likely to be
>> > safer to do that.
>> >
>> > Can you work up a patch?  I've got too many things to juggle right
>> > now, and this is apparently a release concern.
>>
>> For the moment, I decided that it would be better not to propose
>> another solution.  Mark has been presented with a couple of alternatives:
>>
>> 1) Documentation change (ADAC should be same as CC).
>> 2) Port from trunk Geoffrey Keating's configure patch to check if
>>    -Wno-long-long is supported by the ada driver.
>
> I'm in favor of (2), for what it's worth; we know it works on the
> trunk.

Do you have a pointer to (2)?

Thanks,

--
Mark Mitchell                   mark@codesourcery.com
CodeSourcery, LLC               http://www.codesourcery.com

Re: 3.1 branch ada bootstrap fails

[Zack Weinberg]

On Mon, May 06, 2002 at 11:12:35AM -0400, John David Anglin wrote:
> > I am not sure whether ALL_CFLAGS wouldn't have other junk, such as
> > unwanted -D switches.  However, if it doesn't, it is likely to be
> > safer to do that.
> > 
> > Can you work up a patch?  I've got too many things to juggle right
> > now, and this is apparently a release concern.
> 
> For the moment, I decided that it would be better not to propose
> another solution.  Mark has been presented with a couple of alternatives:
> 
> 1) Documentation change (ADAC should be same as CC).
> 2) Port from trunk Geoffrey Keating's configure patch to check if
>    -Wno-long-long is supported by the ada driver.

I'm in favor of (2), for what it's worth; we know it works on the
trunk.

zw

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> I am not sure whether ALL_CFLAGS wouldn't have other junk, such as
> unwanted -D switches.  However, if it doesn't, it is likely to be
> safer to do that.
> 
> Can you work up a patch?  I've got too many things to juggle right
> now, and this is apparently a release concern.

For the moment, I decided that it would be better not to propose
another solution.  Mark has been presented with a couple of alternatives:

1) Documentation change (ADAC should be same as CC).
2) Port from trunk Geoffrey Keating's configure patch to check if
   -Wno-long-long is supported by the ada driver.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: 3.1 branch ada bootstrap fails

[Zack Weinberg]

On Fri, May 03, 2002 at 02:03:21PM -0400, John David Anglin wrote:
> > The sed needs another operation, -e 's/-Wno-long-long//g', but that
> > isn't the immediate problem.  Since you've got all three of -pedantic
> > -Wtraditional -Wno-long-long on your command line, this is not working
> > at all.  Up above, we have
> > 
> > ALL_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(ADAFLAGS)
> > MOST_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(SOME_ADAFLAGS)
> > 
> > Perhaps this should be $(LOOSE_CFLAGS) $(ADA_CFLAGS) ?
> 
> I think it should be $(ALL_CFLAGS) $(ADA_CFLAGS) ...  It has the same
> processing and includes X_CFLAGS and T_CFLAGS.  Didn't want to rock
> the boat too much when I proposed the patch originally.  The trunk
> has the same problem.

I am not sure whether ALL_CFLAGS wouldn't have other junk, such as
unwanted -D switches.  However, if it doesn't, it is likely to be
safer to do that.

Can you work up a patch?  I've got too many things to juggle right
now, and this is apparently a release concern.

zw

Re: 3.1 branch ada bootstrap fails

[Alexandre Oliva]

On May  3, 2002, Geoff Keating <geoffk@geoffk.org> wrote:

> Alexandre Oliva <aoliva@redhat.com> writes:
>> On May  3, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
>> 
>> > I believe that you have an incompatibility between the GCC driver used
>> > for your build and gnatgcc.  There is this configure check:
>> 
>> > checking whether gcc accepts -Wno-long-long... yes
>> 
>> This tests whether gcc accepts this flag.  It doesn't check whether
>> gnatgcc accept it, nor whether it works when compiling an Ada source
>> file (as opposed to a C source file).

> 2002-03-14  Geoffrey Keating  <geoffk@redhat.com>

>         * configure.in: Don't pass -Wno-long-long to a ADA compiler
>         that doesn't support it.
>         * configure: Regenerate.

Most definitely.  Mark, care to approve this for 3.1?

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

Re: 3.1 branch ada bootstrap fails

[Geoff Keating]

Alexandre Oliva <aoliva@redhat.com> writes:

> On May  3, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
> 
> > I believe that you have an incompatibility between the GCC driver used
> > for your build and gnatgcc.  There is this configure check:
> 
> > checking whether gcc accepts -Wno-long-long... yes
> 
> This tests whether gcc accepts this flag.  It doesn't check whether
> gnatgcc accept it, nor whether it works when compiling an Ada source
> file (as opposed to a C source file).

This is different in the mainline.  Perhaps my patch that changed
this should be imported to the branch?

2002-03-14  Geoffrey Keating  <geoffk@redhat.com>

        * configure.in: Don't pass -Wno-long-long to a ADA compiler
        that doesn't support it.
        * configure: Regenerate.

-- 
- Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> On May  3, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:
> 
> > I believe that you have an incompatibility between the GCC driver used
> > for your build and gnatgcc.  There is this configure check:
> 
> > checking whether gcc accepts -Wno-long-long... yes
> 
> This tests whether gcc accepts this flag.  It doesn't check whether
> gnatgcc accept it, nor whether it works when compiling an Ada source
> file (as opposed to a C source file).

Although you can't tell from the report, it looks as if two different
compiler drivers were used for the build (ie., -Wno-long-long wasn't
in the documented_lang_options struct of gnatgcc).  It appears that
documented language options are normally ignored unless extra_warnings
is set.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: 3.1 branch ada bootstrap fails

[Alexandre Oliva]

On May  3, 2002, "John David Anglin" <dave@hiauly1.hia.nrc.ca> wrote:

> I believe that you have an incompatibility between the GCC driver used
> for your build and gnatgcc.  There is this configure check:

> checking whether gcc accepts -Wno-long-long... yes

This tests whether gcc accepts this flag.  It doesn't check whether
gnatgcc accept it, nor whether it works when compiling an Ada source
file (as opposed to a C source file).

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

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> The sed needs another operation, -e 's/-Wno-long-long//g', but that
> isn't the immediate problem.  Since you've got all three of -pedantic
> -Wtraditional -Wno-long-long on your command line, this is not working
> at all.  Up above, we have
> 
> ALL_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(ADAFLAGS)
> MOST_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(SOME_ADAFLAGS)
> 
> Perhaps this should be $(LOOSE_CFLAGS) $(ADA_CFLAGS) ?

I think it should be $(ALL_CFLAGS) $(ADA_CFLAGS) ...  It has the same
processing and includes X_CFLAGS and T_CFLAGS.  Didn't want to rock
the boat too much when I proposed the patch originally.  The trunk
has the same problem.

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

Re: 3.1 branch ada bootstrap fails

[Zack Weinberg]

On Fri, May 03, 2002 at 01:09:47PM -0400, John David Anglin wrote:
> > 2002-05-02  John David Anglin  <dave@hiauly1.hia.nrc.ca>
> > 
> >	* Makefile.in (ALL_ADAFLAGS, MOST_ADAFLAGS): Add CFLAGS to defines.
> 
> I don't think that the patch is the cause of the problem:
> 
> gcc -c -g  -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wt
> raditional -pedantic -Wno-long-long -mdisable-indexing   -W -Wall -gnatpg -gnata
>  -I- -I. -I../../../gcc/gcc/ada ../../../gcc/gcc/ada/ada.ads
> 
> does not cause an error in my builds and CFLAGS has always been used in
> compiling ada files, except for a couple of days.

However, -pedantic -Wno-long-long are not supposed to be applied to
front ends other than C.  There is logic in ada/Makefile.in intended
to strip them:

# This is the variable actually used when we compile.
LOOSE_CFLAGS = `echo $(CFLAGS) $(WARN2_CFLAGS)|sed -e 's/-pedantic//g' -e 's/-Wtraditional//g'`
ALL_CFLAGS = $(INTERNAL_CFLAGS) $(X_CFLAGS) $(T_CFLAGS) $(LOOSE_CFLAGS) \
	$(XCFLAGS)

The sed needs another operation, -e 's/-Wno-long-long//g', but that
isn't the immediate problem.  Since you've got all three of -pedantic
-Wtraditional -Wno-long-long on your command line, this is not working
at all.  Up above, we have

ALL_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(ADAFLAGS)
MOST_ADAFLAGS = $(CFLAGS) $(ADA_CFLAGS) $(SOME_ADAFLAGS)

Perhaps this should be $(LOOSE_CFLAGS) $(ADA_CFLAGS) ?

zw

Re: 3.1 branch ada bootstrap fails

[John David Anglin]

> 2002-05-02  John David Anglin  <dave@hiauly1.hia.nrc.ca>
> 
>	* Makefile.in (ALL_ADAFLAGS, MOST_ADAFLAGS): Add CFLAGS to defines.

I don't think that the patch is the cause of the problem:

gcc -c -g  -W -Wall -Wwrite-strings -Wstrict-prototypes -Wmissing-prototypes -Wt
raditional -pedantic -Wno-long-long -mdisable-indexing   -W -Wall -gnatpg -gnata
 -I- -I. -I../../../gcc/gcc/ada ../../../gcc/gcc/ada/ada.ads

does not cause an error in my builds and CFLAGS has always been used in
compiling ada files, except for a couple of days.

> This change causes ada to fail as of today as follows:
> 
> gnatgcc -c -g  -W -Wall -Wwrite-strings -Wstrict-prototypes
> -Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long    -W -Wall
> -gnatpg -gnata -I- -I. -I/usr/src/cvs/gcc-3.1/
> gcc/ada /usr/src/cvs/gcc-3.1/gcc/ada/ada.ads
> gnat1: Invalid option `-Wno-long-long'

I believe that you have an incompatibility between the GCC driver used
for your build and gnatgcc.  There is this configure check:

checking whether gcc accepts -Wno-long-long... yes

Dave
-- 
J. David Anglin                                  dave.anglin@nrc.ca
National Research Council of Canada              (613) 990-0752 (FAX: 952-6605)

3.1 branch ada bootstrap fails

[Gordon Sadler]

2002-05-02  John David Anglin  <dave@hiauly1.hia.nrc.ca>

        * Makefile.in (ALL_ADAFLAGS, MOST_ADAFLAGS): Add CFLAGS to defines.

This change causes ada to fail as of today as follows:

gnatgcc -c -g  -W -Wall -Wwrite-strings -Wstrict-prototypes
-Wmissing-prototypes -Wtraditional -pedantic -Wno-long-long    -W -Wall
-gnatpg -gnata -I- -I. -I/usr/src/cvs/gcc-3.1/
gcc/ada /usr/src/cvs/gcc-3.1/gcc/ada/ada.ads
gnat1: Invalid option `-Wno-long-long'
make[3]: *** [ada.o] Error 1


-- 
Gordon Sadler