bootstrapping woes on Windows, a compilation

bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

(Except point 5, the others have already been reported; I'm just
collecting them so they're all in one place)

Bootstrapping the trunk on a Spanish Edition Windows XP, with gcc
(GCC) 4.2.1-sjlj (mingw32-2).

1.- (Already explained by Jason)  The leim/CXTERM-DIC dir, or at least
the .tit files, must be checked out with -kb or the build fails. That
will have to be added to nt/INSTALL (or fixed in the CVS?).

2.- (Already reported by Eric Lilja)  Compiling the .el files gives a
ton of warnings

  Warning: Default coding system `iso-latin-1' disagrees with
  system codeset `cp1252' for this locale.

which disappear if I do "set LANG=en_US.UTF-8". Once built, if LANG is
not set, I get the same warning on *Messages* on starting Emacs.

3.- (Already reported by Zhang Wei)  I *must* use
--enable-font-backend, or the bootstrap dies:

oo-spd/i386/temacs0.a(emacs.o): In function `main':
C:/emacs/src/emacs.c:1616: undefined reference to `syms_of_font'
oo-spd/i386/temacs1.a(fontset.o): In function `load_font_get_repertory':
C:/emacs/src/fontset.c:514: undefined reference to `Ffont_get'
oo-spd/i386/temacs1.a(fontset.o): In function `Fset_fontset_font':
C:/emacs/src/fontset.c:1569: undefined reference to `QCregistry'
C:/emacs/src/fontset.c:1571: undefined reference to `Ffont_spec'
C:/emacs/src/fontset.c:1442: undefined reference to `Ffont_spec'
C:/emacs/src/fontset.c:1447: undefined reference to `font_parse_xlfd'
C:/emacs/src/fontset.c:1452: undefined reference to `font_unparse_xlfd'
C:/emacs/src/fontset.c:1579: undefined reference to `Ffont_spec'
C:/emacs/src/fontset.c:1442: undefined reference to `Ffont_spec'
C:/emacs/src/fontset.c:1447: undefined reference to `font_parse_xlfd'
C:/emacs/src/fontset.c:1452: undefined reference to `font_unparse_xlfd'
collect2: ld returned 1 exit status
make[2]: *** [oo-spd/i386/temacs.exe] Error 1

4.- (Already reported by Óscar Fuentes and partially fixed by Jason)
The generated files containing character data are missing:

make[1]: *** No rule to make target `../lisp/international/charprop.el', needed
by `DOC'.  Stop.
make[1]: Leaving directory `T:/emacscvs/src/emacs/lib-src'
make: *** [bootstrap-gmake] Error 2

They are generated now, but not in time (Jason said), so the second
bootstrap can succeed. (Which means, I suppose, that we'll have to add
a realclean or distclean target to lisp/Makefile.w32-in to remove
these generated files.)

5.- If I use "-DENABLE_CHECKING=1 -DSITELOAD_PURESIZE_EXTRA=200000
-fno-crossjumping", Emacs crashes:

"./oo-spd/i386/temacs.exe" -batch -l loadup bootstrap
make[1]: *** [bootstrap-emacs] Error -2147483645

The output of DrMingw is attached.

Other than these, Emacs bootstraps the second time around. But, as
Jason said, the size of the default font is all wrong.


             Juanma



temacs.exe caused a Breakpoint at location 7c911230 in module ntdll.dll.

Registers:
eax=00000006 ebx=00000001 ecx=7c92056d edx=00040002 esi=00000000 edi=01c26000
eip=7c911230 esp=0082fe3c ebp=0082fe58 iopl=0         nv up ei pl zr na po nc
cs=001b  ss=0023  ds=0023  es=0023  fs=003b  gs=0000             efl=00000246

Call stack:
AddrPC     AddrReturn AddrFrame  AddrStack  Params
7C911230   0101F16B   0082FE58   0082FE3C   77C2FCC0   016EAE70
01761AC0   0000006B
7C911230  ntdll.dll:7C911230  DbgBreakPoint
0101F16B   01135BE2   0082FE78   0082FE3C   017627F8   01761AC0
0000006B   01C1F861
0101F16B  temacs.exe:0101F16B  free_misc  alloc.c:3298
	...
	free_misc (misc)
	     Lisp_Object misc;
>	{
	  XMISCTYPE (misc) = Lisp_Misc_Free;
	  XMISC (misc)->u_free.chain = marker_free_list;
	...

01135BE2   01135F0B   0082FE98   0082FE3C   00000220   01C1CC05
00000000   01C1CC05
01135BE2  temacs.exe:01135BE2  make_sub_char_table  chartab.c:105
	...
	  table = Fmake_vector (make_number (size), defalt);
	  XSETPVECTYPE (XVECTOR (table), PVEC_SUB_CHAR_TABLE);
>	  XSUB_CHAR_TABLE (table)->depth = make_number (depth);
	  XSUB_CHAR_TABLE (table)->min_char = make_number (min_char);
	  XSETSUB_CHAR_TABLE (table, XSUB_CHAR_TABLE (table));
	...

01135F0B   010F186A   0082FEB8   0082FE3C   01C26004   00000000
01C1CC0D   0101FA82
01135F0B  temacs.exe:01135F0B  char_table_set  chartab.c:420
	...
	      if (! SUB_CHAR_TABLE_P (sub))
	{
>	  sub = make_sub_char_table (1, i * chartab_chars[0], sub);
	  tbl->contents[i] = sub;
	}
	...

010F186A   01002CDA   0082FEF8   0082FE3C   016DF46F   00000003
00000000   0082FF4C
010F186A  temacs.exe:010F186A  init_syntax_once  syntax.c:3354
	...
	  /* Control characters should not be whitespace.  */
	  temp = XVECTOR (Vsyntax_code_object)->contents[(int) Spunct];
>	  for (i = 0; i <= ' ' - 1; i++)
	    SET_RAW_SYNTAX_ENTRY (Vstandard_syntax_table, i, temp);
	  SET_RAW_SYNTAX_ENTRY (Vstandard_syntax_table, 0177, temp);
	...

01002CDA   0100124B   0082FF68   0082FE3C   7FFDF000   01C8659E
0082FFB0   0100124B
01002CDA  temacs.exe:01002CDA  main  emacs.c:1293
	...
	      init_coding_once ();
	      init_syntax_once ();/* Create standard syntax table.  */
>	      init_category_once ();/* Create standard category table.  */
	      /* Must be done before init_buffer.  */
	      init_casetab_once ();
	...

0100124B   01001298   0082FFB0   0082FE3C   00000001   00000009
0082FFF0   7C816FD7
0100124B  temacs.exe:0100124BC:\bin\emacs\git\src\oo-spd\i386\temacs.exe:
No symbol found

01001298   7C816FD7   0082FFC0   0082FE3C   E399EB14   01C8659E
7FFDF000   80000003
01001298  temacs.exe:01001298C:\bin\emacs\git\src\oo-spd\i386\temacs.exe:
No symbol found

7C816FD7   00000000   0082FFF0   0082FE3C   0100486B   00000000
78746341   00000020
7C816FD7  kernel32.dll:7C816FD7  RegisterWaitForInputIdle
DEBUG_EVENT:
	dwDebugEventCode = EXIT_PROCESS_DEBUG_EVENT
	dwProcessId = AF8
	dwThreadId = A14
	dwExitCode = 80000003

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 14:52:23 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> 
> 4.- (Already reported by Óscar Fuentes and partially fixed by Jason)
> The generated files containing character data are missing:
> 
> make[1]: *** No rule to make target `../lisp/international/charprop.el', needed
> by `DOC'.  Stop.
> make[1]: Leaving directory `T:/emacscvs/src/emacs/lib-src'
> make: *** [bootstrap-gmake] Error 2

This should be fixed now; please try.  I only tested the bootstrap
with GNU Make and a Unixy shell, so please test the other supported
combinations as well.

> They are generated now, but not in time (Jason said), so the second
> bootstrap can succeed. (Which means, I suppose, that we'll have to add
> a realclean or distclean target to lisp/Makefile.w32-in to remove
> these generated files.)

I will look into this now.

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 14:52:23 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> 
> They are generated now, but not in time (Jason said), so the second
> bootstrap can succeed. (Which means, I suppose, that we'll have to add
> a realclean or distclean target to lisp/Makefile.w32-in to remove
> these generated files.)

I don't think we need to add any distclean targets for this.  These
files are generated only during bootstrap, so only bootstrap-clean
targets should remove them.  And the changes I committed just now to
nt/makefile.w32-in already regenerate those files every bootstrap.

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 4:10 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> This should be fixed now; please try.  I only tested the bootstrap
> with GNU Make and a Unixy shell, so please test the other supported
> combinations as well.

I'm testing with

 gcc (GCC) 4.2.1-sjlj (mingw32-2)
 GNU Make 3.80, sed, etc. from UnxUtils
 CMD

I get:

make   unidatagen-CMD
make[1]: Entering directory `C:/emacs/nt'
if exist ..\admin\unidata\UnicodeData.txt \
   make -w  -C ../admin/unidata
make[2]: Entering directory `C:/emacs/admin/unidata'
"../../src/oo-spd/i386/emacs.exe" -Q --multibyte -batch -f
batch-byte-compile unidata-gen.el
Wrote c:/emacs/admin/unidata/unidata-gen.elc
sed -e "s/\([^;]*\);\(.*\)/(#x\1 \"\2\")/" -e "s/;/\" \"/g" <
UnicodeData.txt > unidata.txt
make[2]: Leaving directory `C:/emacs/admin/unidata'
makefile:341: *** unterminated variable reference.  Stop.
make[1]: *** [unidatagen-CMD] Error 2

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Angelo Graziosi]

Juanma Barranquero wrote:

 > I'm just collecting them so they're all in one place


Could we add also this 
http://lists.gnu.org/archive/html/emacs-devel/2008-02/msg00052.html, 
even if it happens on Cygwin?


Thanks,
    Angelo.

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 4:18 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> I don't think we need to add any distclean targets for this.  These
> files are generated only during bootstrap, so only bootstrap-clean
> targets should remove them.

I though distclean was supposed to leave the source tree free of
generated files.

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 4:58 PM, Angelo Graziosi <angelo.graziosi@alice.it> wrote:

> Could we add also this
> http://lists.gnu.org/archive/html/emacs-devel/2008-02/msg00052.html,
> even if it happens on Cygwin?

Be my guest.

(I mean, just update the message and post it again...)

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 16:58:46 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> On Feb 2, 2008 4:18 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > I don't think we need to add any distclean targets for this.  These
> > files are generated only during bootstrap, so only bootstrap-clean
> > targets should remove them.
> 
> I though distclean was supposed to leave the source tree free of
> generated files.

It is supposed to remove files generated during the normal
(i.e. non-CVS) build.  In other words, files that are not in the
official tarballs.  But these files _will_ be present in an official
distribution, not generated by "make", so they should not be removed
by distclean.

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 16:57:55 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> makefile:341: *** unterminated variable reference.  Stop.
> make[1]: *** [unidatagen-CMD] Error 2

Shame on me.  Should be fixed now.

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 5:24 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> > Date: Sat, 2 Feb 2008 16:58:46 +0100

> It is supposed to remove files generated during the normal
> (i.e. non-CVS) build.  In other words, files that are not in the
> official tarballs.

Aha, I understand now.

Thanks,

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 5:35 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> Shame on me.

No, shame on *me*. I should've taken a look at the makefile and see
that it was simply a missing parenthesis, but I was a bit distracted
(I'm officially not at the keyboard right now ;-)

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 5:35 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> Should be fixed now.

Now the cleaning worked, and the uni-* and charprop files were
generated, but they stay in admin/unidata.

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 5:35 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> Should be fixed now.

Now the cleaning worked, and the uni-* and charprop files were
generated, but they stay in admin/unidata.

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 18:25:20 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> On Feb 2, 2008 5:35 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > Should be fixed now.
> 
> Now the cleaning worked, and the uni-* and charprop files were
> generated, but they stay in admin/unidata.

That's a real gotcha.  Please try again.

Re: bootstrapping woes on Windows, a compilation

[Juanma Barranquero]

On Feb 2, 2008 6:56 PM, Eli Zaretskii <eliz@gnu.org> wrote:

> That's a real gotcha.  Please try again.

Success! Thanks.

             Juanma

Re: bootstrapping woes on Windows, a compilation

[Eli Zaretskii]

> Date: Sat, 2 Feb 2008 20:13:59 +0100
> From: "Juanma Barranquero" <lekktu@gmail.com>
> Cc: emacs-devel@gnu.org
> 
> On Feb 2, 2008 6:56 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > That's a real gotcha.  Please try again.
> 
> Success! Thanks.

Thanks for testing.