bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

I've recently tried to build current emacs versions on IRIX 6.5
(6.5.30f, actually), but always failed in the same way.

The latest attempt was like this:

* Configure emacs 24.0.90 with

  /vol/src/gnu/emacs/emacs-24.0.90/configure --prefix=/vol/gnu --infodir=/vol/gnu/info --mandir=/vol/gnu/man --libexecdir=/vol/gnu/lib --localstatedir=/vol/gnu/share/emacs/var --with-xpm=no --with-jpeg=no --with-png=no --with-gif=no --with-tiff=no

* Build it with gcc 4.6.0 with make -j8.

  The C part of the build complete successfully, but then it fails like
  this:

Dumping under the name emacs
41988 pure bytes used
cd ../lisp; make -w --jobserver-fds=3,4 - --jobserver-fds=3,4 -j compile-first EMACS=/vol/obj/gnu/emacs/emacs-24.0.90/src/bootstrap-emacs
make[2]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
make[2]: Nothing to be done for `compile-first'.
make[2]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
cd ../lisp; make -w --jobserver-fds=3,4 - --jobserver-fds=3,4 -j autoloads EMACS=/vol/obj/gnu/emacs/emacs-24.0.90/src/bootstrap-emacs
make[2]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
Compiling /vol/src/gnu/emacs/emacs-24.0.90/src/../lisp/emacs-lisp/backquote.el
emacs: Invalid function: make-local-variable
make[2]: *** [compile-onefile] Error 1
make[2]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'

I have no idea what might cause this; emacs 23.2 built fine and mostly
works.  For emacs 23.3, I tried several variations:

* build with gcc 4.4.2

* build with gcc -g3 -O0 instead of the default -g -O2

* build with SGI MIPSpro cc 7.4.4m

None of those makes a difference.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Eli Zaretskii]

> From: Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
> Date: Thu, 06 Oct 2011 16:19:30 +0200
> 
> I've recently tried to build current emacs versions on IRIX 6.5
> (6.5.30f, actually), but always failed in the same way.
> 
> The latest attempt was like this:
> 
> * Configure emacs 24.0.90 with
> 
>   /vol/src/gnu/emacs/emacs-24.0.90/configure --prefix=/vol/gnu --infodir=/vol/gnu/info --mandir=/vol/gnu/man --libexecdir=/vol/gnu/lib --localstatedir=/vol/gnu/share/emacs/var --with-xpm=no --with-jpeg=no --with-png=no --with-gif=no --with-tiff=no
> 
> * Build it with gcc 4.6.0 with make -j8.
> 
>   The C part of the build complete successfully, but then it fails like
>   this:

Does it help to drop the -j8 switch to Make?

And what version of GNU Make is that, btw?

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Glenn Morris]

If you have an SGI IRIX, maybe you could help out with
http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7643 ?

The original submitter never replied.

Though presumably you don't see that problem since the C build completed
for you (also I'm not sure if you are using the same kind of system).

Thanks!

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Eli Zaretskii <eliz@gnu.org> writes:

> Does it help to drop the -j8 switch to Make?

Unfortunately not: just running make yields

Loading tooltip...
Finding pointers to doc strings...
Finding pointers to doc strings...done
Dumping under the name emacs
1465234 pure bytes used
Adding name emacs-24.0.90.1
emacs: Invalid function: make-local-variable
make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/src'
cd lisp; make all                            \
          CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
          LDFLAGS='-Wl,-znocombreloc ' MAKE='make'
make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
make[2]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
make[2]: Nothing to be done for `compile-targets'.
make[2]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
(export PARALLEL; PARALLEL=0; cd leim; make all  \
          CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
          LDFLAGS='-Wl,-znocombreloc ' MAKE='make')
make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/leim'
mkdir quail
touch stamp-subdir
EMACSLOADPATH=/vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp LC_ALL=C ../src/emacs -batch --no-site-file --no-site-lisp -l /vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp/international/titdic-cnv \
          -f batch-titdic-convert -dir quail /vol/src/gnu/emacs/emacs-24.0.90/leim/CXTERM-DIC; \
          echo "changed" > changed.tit
emacs: Invalid function: make-local-variable
Compiling quail/CCDOSPY.el
emacs: Invalid function: make-local-variable
make[1]: *** [quail/CCDOSPY.elc] Error 1
make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/leim'
make: *** [leim] Error 2

> And what version of GNU Make is that, btw?

ro@columba 150 > make --version
GNU Make 3.81

The same one that worked for 23.2 and all my GCC bootstraps.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#7643: bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

[Cc'ing to the correct bug.]

Glenn Morris <rgm@gnu.org> writes:

> If you have an SGI IRIX, maybe you could help out with
> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=7643 ?
>
> The original submitter never replied.

I can at least try...

> Though presumably you don't see that problem since the C build completed
> for you (also I'm not sure if you are using the same kind of system).

I fear that's exactly the problem: IPv6 support was only added in a
later update of IRIX 6.5 (don't recall exactly which one), so without
knowing what Nelson is (was) running, it's hard to tell.  Maybe emacs
assumes IPv6 support unconditionally (or the test was insufficient)?

I did have various problems building 23.3 on 6.5.30 myself, but nothing
related to IPv6.  I didn't try with cc, though.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Eli Zaretskii]

> From: Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
> Cc: 9684@debbugs.gnu.org
> Date: Fri, 07 Oct 2011 11:11:46 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Does it help to drop the -j8 switch to Make?
> 
> Unfortunately not: just running make yields
> 
> Loading tooltip...
> Finding pointers to doc strings...
> Finding pointers to doc strings...done
> Dumping under the name emacs
> 1465234 pure bytes used
> Adding name emacs-24.0.90.1
> emacs: Invalid function: make-local-variable
> make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/src'
> cd lisp; make all                            \
>           CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
>           LDFLAGS='-Wl,-znocombreloc ' MAKE='make'
> make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[2]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[2]: Nothing to be done for `compile-targets'.
> make[2]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> (export PARALLEL; PARALLEL=0; cd leim; make all  \
>           CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
>           LDFLAGS='-Wl,-znocombreloc ' MAKE='make')
> make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/leim'
> mkdir quail
> touch stamp-subdir
> EMACSLOADPATH=/vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp LC_ALL=C ../src/emacs -batch --no-site-file --no-site-lisp -l /vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp/international/titdic-cnv \
>           -f batch-titdic-convert -dir quail /vol/src/gnu/emacs/emacs-24.0.90/leim/CXTERM-DIC; \
>           echo "changed" > changed.tit
> emacs: Invalid function: make-local-variable
> Compiling quail/CCDOSPY.el
> emacs: Invalid function: make-local-variable
> make[1]: *** [quail/CCDOSPY.elc] Error 1

Can you run any of these 2 commands under a debugger, put a breakpoint
in a function called xsignal1, and when it breaks, show the backtrace?
Preferably do this with GDB, and start GDB from the src directory, so
it reads the .gdbinit file there.  Also, please recompile Emacs
without the -O2 optimization switch, so that information shown by the
debugger would be reliable.

Thanks.

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Eli Zaretskii <eliz@gnu.org> writes:

> Can you run any of these 2 commands under a debugger, put a breakpoint
> in a function called xsignal1, and when it breaks, show the backtrace?

Sure:

(gdb) set environment EMACSLOADPATH /vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp
(gdb) b xsignal1
Breakpoint 3 at 0x10294d2c
(gdb) run -batch --no-site-file --no-site-lisp -f batch-byte-compile quail/CCDOSPY.el
Starting program: /vol/obj/gnu/emacs/emacs-24.0.90/src/emacs -batch --no-site-file --no-site-lisp -f batch-byte-compile quail/CCDOSPY.el

Breakpoint 3, xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
(gdb) where
#0  xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
#1  0x10298f1c in Ffuncall (nargs=2, args=0x7ffb7a10) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3029
#2  0x10306074 in exec_byte_code (bytestr=272696025, vector=272696045, maxdepth=8, args_template=0, nargs=0, args=0x7ffb7d14) at /vol/src/gnu/emacs/emacs-24.0.90/src/bytecode.c:785
#3  0x1029963c in funcall_lambda (fun=272695997, nargs=0, arg_vector=0x7ffb7d14) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3139
#4  0x10298ea8 in Ffuncall (nargs=1, args=0x7ffb7d10) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3023
#5  0x10297ea8 in call0 (fn=274434914) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:2728
#6  0x1020eed4 in get_minibuffer (depth=0) at /vol/src/gnu/emacs/emacs-24.0.90/src/minibuf.c:815
#7  0x101fc68c in init_buffer () at /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.c:5098
#8  0x101b5520 in main (argc=7, argv=0x7ffb7ef4) at /vol/src/gnu/emacs/emacs-24.0.90/src/emacs.c:1452

Lisp Backtrace:
"make-local-variable" (0x7ffb7a14)
"minibuffer-inactive-mode" (0x7ffb7d14)

> Preferably do this with GDB, and start GDB from the src directory, so
> it reads the .gdbinit file there.  Also, please recompile Emacs
> without the -O2 optimization switch, so that information shown by the
> debugger would be reliable.

Of course: recompiled with -g3 -O0.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Eli Zaretskii]

> From: Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
> Cc: 9684@debbugs.gnu.org
> Date: Fri, 07 Oct 2011 15:40:56 +0200
> 
> Breakpoint 3, xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
> (gdb) where
> #0  xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
> #1  0x10298f1c in Ffuncall (nargs=2, args=0x7ffb7a10) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3029
> #2  0x10306074 in exec_byte_code (bytestr=272696025, vector=272696045, maxdepth=8, args_template=0, nargs=0, args=0x7ffb7d14) at /vol/src/gnu/emacs/emacs-24.0.90/src/bytecode.c:785
> #3  0x1029963c in funcall_lambda (fun=272695997, nargs=0, arg_vector=0x7ffb7d14) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3139

Please show what these commands produce:

 (gdb) frame 3
 (gdb) pp fun
 (gdb) frame 1
 (gdb) pp args[0]
 (gdb) pp args[1]

> Of course: recompiled with -g3 -O0.

Thanks.

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Eli Zaretskii <eliz@gnu.org> writes:

> Please show what these commands produce:
>
>  (gdb) frame 3
>  (gdb) pp fun
>  (gdb) frame 1
>  (gdb) pp args[0]
>  (gdb) pp args[1]

Unfortunately, they don't really work:

Breakpoint 3, xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
(gdb) frame 3
#3  0x1029963c in funcall_lambda (fun=272695997, nargs=0, arg_vector=0x7ffb7d14) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3139
(gdb) pp fun
Address of symbol "print_output_debug_flag" is unknown.
(gdb) frame 1
#1  0x10298f1c in Ffuncall (nargs=2, args=0x7ffb7a10) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3029
(gdb) pp args[0]
Address of symbol "print_output_debug_flag" is unknown.
(gdb)pp args[1] 
Address of symbol "print_output_debug_flag" is unknown.

This is really strange, though:

ro@columba 221 > nm print.o|grep print_output_debug_flag
[44]    |         0|       4|OBJT |GLOB |DEFAULT  |3      |print_output_debug_flag
ro@columba 222 > nm emacs|grep print_output_debug_flag
[2868]  | 273886352|       4|OBJT |GLOB |DEFAULT  |MIPS_SCOMMON|print_output_debug_flag

Since the save/restore of print_output_debug_flag seems to be for the
benefit of w32 only, I've tried the following instead:

Breakpoint 3, xsignal1 (error_symbol=274352730, arg=274355066) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:1773
(gdb) dis 3
(gdb) frame 3
#3  0x1029963c in funcall_lambda (fun=272695997, nargs=0, arg_vector=0x7ffb7d34) at /vol/src/gnu/emacs/emacs-24.0.90/src/eval.c:3139
(gdb) set $tmp = fun
(gdb) set safe_debug_print ($tmp)
emacs: Invalid function: external-debugging-output
[Inferior 1 (process 24041591) exited with code 01]
The program being debugged exited while in a function called from GDB.
Evaluation of the expression containing the function
(safe_debug_print) will be abandoned.
(gdb) frame 1
No registers.

Seriously ugly.  This is all gdb 7.3, so nothing newer to try (except
perhaps for gdb mainline).

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Dan Nicolaescu]

Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>> Does it help to drop the -j8 switch to Make?
>
> Unfortunately not: just running make yields
>
> Loading tooltip...
> Finding pointers to doc strings...
> Finding pointers to doc strings...done
> Dumping under the name emacs
> 1465234 pure bytes used
> Adding name emacs-24.0.90.1
> emacs: Invalid function: make-local-variable
> make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/src'
> cd lisp; make all                            \
>           CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
>           LDFLAGS='-Wl,-znocombreloc ' MAKE='make'
> make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[2]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[2]: Nothing to be done for `compile-targets'.
> make[2]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> make[1]: Leaving directory `/vol/obj/gnu/emacs/emacs-24.0.90/lisp'
> (export PARALLEL; PARALLEL=0; cd leim; make all  \
>           CC='gcc -std=gnu99' CFLAGS='-g -O2' CPPFLAGS='' \
>           LDFLAGS='-Wl,-znocombreloc ' MAKE='make')
> make[1]: Entering directory `/vol/obj/gnu/emacs/emacs-24.0.90/leim'
> mkdir quail
> touch stamp-subdir
> EMACSLOADPATH=/vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp LC_ALL=C ../src/emacs -batch --no-site-file --no-site-lisp -l /vol/src/gnu/emacs/emacs-24.0.90/leim/../lisp/international/titdic-cnv \
>           -f batch-titdic-convert -dir quail /vol/src/gnu/emacs/emacs-24.0.90/leim/CXTERM-DIC; \
>           echo "changed" > changed.tit
> emacs: Invalid function: make-local-variable

Maybe there's a problem with dumping, can you try running the above
command and using the  ../src/temacs binary instead of ../src/emacs ? 

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Eli Zaretskii]

> From: Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE>
> Cc: 9684@debbugs.gnu.org
> Date: Fri, 07 Oct 2011 17:57:33 +0200
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Please show what these commands produce:
> >
> >  (gdb) frame 3
> >  (gdb) pp fun
> >  (gdb) frame 1
> >  (gdb) pp args[0]
> >  (gdb) pp args[1]
> 
> Unfortunately, they don't really work:

Then you will have to do it the hard way:

 (gdb) p fun
 (gdb) xtype

Then use the relevant x* commands, depending on what xtype says.
E.g., if it says that fun is a cons cell, use xcar and xcdr to display
the components, then xtype again for each component, etc.

But before you get down to all this mess, it would be good to use
Dan's suggestion first.

bug#7643: Build failure for SGI IRIX MIPS

[Glenn Morris]

Rainer Orth wrote:

> I fear that's exactly the problem: IPv6 support was only added in a
> later update of IRIX 6.5 (don't recall exactly which one), so without
> knowing what Nelson is (was) running, it's hard to tell.  Maybe emacs
> assumes IPv6 support unconditionally (or the test was insufficient)?

Thanks for looking at this.

However, all the relevant places in process.c seems to be inside #ifdef
AF_INET6, where AF_INET6 is defined by the system headers, not by Emacs.
So if IPv6 support is not present these sections should never be seen.

bug#7643: Build failure for SGI IRIX MIPS

[Glenn Morris]

Web search suggests maybe some IRIXs need -DINET6?

http://www.mail-archive.com/openpkg-users@openpkg.org/msg02606.html

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Paul Eggert]

One possibility is that it's a garbage collector bug.
You might try commenting out this line in src/s/irix6-5.h:

#define GC_MARK_STACK GC_MAKE_GCPROS_NOOPS

and building from scratch.  If this fixes the problem,
we can investigate what went wrong in stack-marking
(though that won't be fun).

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Dan,

> Maybe there's a problem with dumping, can you try running the above
> command and using the  ../src/temacs binary instead of ../src/emacs ? 

yes, using temacs instead works fine (but slooow, of course).

Any suggestions on how to best investigate undump failures?  I'll first
try both SGI MIPSpro cc and older gcc releases (3.4, 4.4) to check if
they make a difference.

Thanks.
        Rainer

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

I've just tried building 24.0.90 with gcc 3.4 and MIPSpro cc:

* With gcc 3.4, the failure is the same as with gcc 4.6.

* With MIPSpro cc, I already get tons of compile failures:

cc-1041 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  The indicated expression must have arithmetic or pointer type.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1053 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A colon is expected at this point.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1168 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  The indicated type conversion is invalid.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1031 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  An integral expression is needed.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1031 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  An integral expression is needed.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1031 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  An integral expression is needed.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1053 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A colon is expected at this point.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1065 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A semicolon is expected at this point.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1055 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A macro invocation has too many arguments.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1055 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A macro invocation has too many arguments.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1054 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  There are not enough arguments in a macro invocation.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1055 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A macro invocation has too many arguments.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1055 cc: ERROR File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  A macro invocation has too many arguments.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

cc-1174 cc: WARNING File = /vol/src/gnu/emacs/emacs-24.0.90/src/buffer.h, Line = 1045
  The variable "PVEC_SUB_CHAR_TABLE" was declared but never referenced.

    Lisp_Object down = CHAR_TABLE_REF (downcase_table, c);
                       ^

I haven't tracked this down yet, and it's certainly another bug.

	Rainer

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Dan Nicolaescu]

Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:

> Dan,
>
>> Maybe there's a problem with dumping, can you try running the above
>> command and using the  ../src/temacs binary instead of ../src/emacs ? 
>
> yes, using temacs instead works fine (but slooow, of course).
>
> Any suggestions on how to best investigate undump failures?  I'll first
> try both SGI MIPSpro cc and older gcc releases (3.4, 4.4) to check if
> they make a difference.

I don't think there have been too many important changes in the dumping
infrastructure, so if something broke is probably an oversight.

You can try compiling a version of emacs that is known to work.
Compile the bzr trunk with the same compiler + flags (maybe use just -O0).
Compare the two "emacs" binaries, use readelf to compare the elf
sections, maybe there's something obvious. 
Compare the two versions of unexelf.o, and maybe the corresponding
preprocessed files (there shouldn't be many interesting differences).
Try linking the trunk temacs with the old unexelf.o and see if you
get better behavior.

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Dan,

> I don't think there have been too many important changes in the dumping
> infrastructure, so if something broke is probably an oversight.
>
> You can try compiling a version of emacs that is known to work.
> Compile the bzr trunk with the same compiler + flags (maybe use just -O0).
> Compare the two "emacs" binaries, use readelf to compare the elf
> sections, maybe there's something obvious. 
> Compare the two versions of unexelf.o, and maybe the corresponding
> preprocessed files (there shouldn't be many interesting differences).
> Try linking the trunk temacs with the old unexelf.o and see if you
> get better behavior.

as an intermediate datapoint: while looking for a version of emacs that
built correctly, I found that even 23.2 didn't when using gcc 4.4 or
3.4.  Fortunately, I had a 9 years old copy of gcc 3.1 around, which was
able to build both a working emacs 23.2 and 24.0.90 (from the tarball).

I'll start digging deeper from here, but it may take a bit.

	Rainer

-- 
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Glenn Morris]

Hi,


Rainer Orth wrote:

> I'll start digging deeper from here, but it may take a bit.

Have you been able to make any progress with this?
We hope to release Emacs 24.3 in the not-too-distant future, and it
would be nice to fix this there, if possible and appropriate.

If you are not able to look into this, it's fine, since you seem to be
approximately the only person using Emacs on IRIX these days, so don't
feel pressured to spends ages on it. :)

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Rainer Orth]

Hi Glenn,

> Have you been able to make any progress with this?
> We hope to release Emacs 24.3 in the not-too-distant future, and it
> would be nice to fix this there, if possible and appropriate.

not really: I've just tried to build emacs 24.2.92 and see similar
successes and failures as with 24.0.90:

* building with gcc 3.1 and MIPSpro 7.4.4m cc produces a working emacs

* building with gcc 3.4 or 4.6 yields the same error as before.

> If you are not able to look into this, it's fine, since you seem to be
> approximately the only person using Emacs on IRIX these days, so don't
> feel pressured to spends ages on it. :)

I suppose I'll be content with having a gcc 3.1-built emacs 24.3 if need
be.  Given that gcc 4.7 was the last release to support IRIX 6.5, I've
not much time for IRIX issues left.

	Rainer

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Paul Eggert]

I recently got access to an IRIX 6.5 host and ported the emacs-24 branch 
to it; see emacs-24 bzr 116961.  So this bug should be partly fixed in 
the next version of Emacs.  I say "partly" because I got it to work with 
IRIX cc but not GCC (some problem with undumping).

http://bzr.savannah.gnu.org/lh/emacs/emacs-24/revision/116961

bug#9684: Emacs 23.3 and 24.0.90 don't build on IRIX 6.5

[Noam Postavsky]

tags 9684 wontfix
close 9684
quit

Rainer Orth <ro@CeBiTec.Uni-Bielefeld.DE> writes:

> I've recently tried to build current emacs versions on IRIX 6.5
> (6.5.30f, actually), but always failed in the same way.

Time to close this as wontfix, since Emacs 26 NEWS has:

    ** Emacs no longer works on IRIX.  We expect that Emacs users are not
    affected by this, as SGI stopped supporting IRIX in December 2013.