Build error with _UNICODE on w32.

Build error with _UNICODE on w32.

[KOBAYASHI Yasuhiro]

I get the following error at w32.c:774.
It was not occured before 9 May 2008.

$ ./configure.bat --with-gcc --no-cygwin --cflags \
  -I/usr/local/include --cflags -D_UNICODE

$ make bootstrap SHELL=bash


echo oo-spd/i386/w32font.o oo-spd/i386/w32uniscribe.o >> oo-spd/i386/buildobj.lst
make[2]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
"./oo-spd/i386/temacs.exe" -batch -l loadup bootstrap
make[1]: *** [bootstrap-emacs] Error 5
make[1]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
make: *** [bootstrap-gmake] Error 2

-- 
KOBAYASHI Yasuhiro <kobayays@gmail.com>

bug#265: Build error with _UNICODE on w32.

[KOBAYASHI Yasuhiro]

I get the following error at w32.c:774.
It was not occured before 9 May 2008.

$ ./configure.bat --with-gcc --no-cygwin --cflags \
  -I/usr/local/include --cflags -D_UNICODE

$ make bootstrap SHELL=bash


echo oo-spd/i386/w32font.o oo-spd/i386/w32uniscribe.o >> oo-spd/i386/buildobj.lst
make[2]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
"./oo-spd/i386/temacs.exe" -batch -l loadup bootstrap
make[1]: *** [bootstrap-emacs] Error 5
make[1]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
make: *** [bootstrap-gmake] Error 2

-- 
KOBAYASHI Yasuhiro <kobayays@gmail.com>

Re: Build error with _UNICODE on w32.

[Eli Zaretskii]

> From: KOBAYASHI Yasuhiro <kobayays@gmail.com>
> Date: Sat, 17 May 2008 13:49:03 +0900
> Cc: 
> 
> I get the following error at w32.c:774.
> It was not occured before 9 May 2008.

What do you mean by ``with _UNICODE''?

> make[2]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
> "./oo-spd/i386/temacs.exe" -batch -l loadup bootstrap
> make[1]: *** [bootstrap-emacs] Error 5

Error code 5 is "Access denied".  Is temacs.exe executable?  That is,
can you invoke it from the shell?

Re: Build error with _UNICODE on w32.

[KOBAYASHI Yasuhiro]

In message Re: Build error with _UNICODE on w32.
 on Sat, 17 May 2008 10:19:31 +0300
 Eli Zaretskii <eliz@gnu.org> wrote:

> > I get the following error at w32.c:774.
> > It was not occured before 9 May 2008.

> What do you mean by ``with _UNICODE''?

It means I execute the following configure.bat.

$ ./configure.bat --with-gcc --no-cygwin --cflags \
  -I/usr/local/include --cflags -D_UNICODE

> > make[2]: Leaving directory `/home/kobayays/src/emacs-vanilla/src'
> > "./oo-spd/i386/temacs.exe" -batch -l loadup bootstrap
> > make[1]: *** [bootstrap-emacs] Error 5

> Error code 5 is "Access denied".  Is temacs.exe executable?  That is,
> can you invoke it from the shell?

No, I gets some error dialog boxes.

-- 
KOBAYASHI Yasuhiro <kobayays@gmail.com>

Re: Build error with _UNICODE on w32.

[Eli Zaretskii]

> From: KOBAYASHI Yasuhiro <kobayays@gmail.com>
> Cc: emacs-pretest-bug@gnu.org
> Date: Sat, 17 May 2008 22:33:33 +0900
> 
> In message Re: Build error with _UNICODE on w32.
>  on Sat, 17 May 2008 10:19:31 +0300
>  Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > > I get the following error at w32.c:774.
> > > It was not occured before 9 May 2008.
> 
> > What do you mean by ``with _UNICODE''?
> 
> It means I execute the following configure.bat.
> 
> $ ./configure.bat --with-gcc --no-cygwin --cflags \
>   -I/usr/local/include --cflags -D_UNICODE

I wasn't aware this is even supported.  It is certainly not
used/tested enough to be recommended for general use.  We don't
consistently use the TCHAR type or any other types that magically
switch to wchar_t under UNICODE compilation, and freely mix text
strings returned from Windows APIs with plain `char *' pointers and
strings.  To get Emacs support the Unicode Windows APIs will take a
major overhaul of the code.

What did you expect to get with the -D_UNICODE switch that you don't
get without it?

Re: Build error with _UNICODE on w32.

[KOBAYASHI Yasuhiro]

In message Re: Build error with _UNICODE on w32.
 on Sat, 17 May 2008 17:01:22 +0300
 Eli Zaretskii <eliz@gnu.org> wrote:

> I wasn't aware this is even supported.  It is certainly not
> used/tested enough to be recommended for general use.  We don't
> consistently use the TCHAR type or any other types that magically
> switch to wchar_t under UNICODE compilation, and freely mix text
> strings returned from Windows APIs with plain `char *' pointers and
> strings.  To get Emacs support the Unicode Windows APIs will take a
> major overhaul of the code.

> What did you expect to get with the -D_UNICODE switch that you don't
> get without it?

I was trying to maintain the IME-patch on w32. I need it to
input the unicode character.

The following is ad hoc one.

w32.c:
@@ -773,7 +773,7 @@
 	      dflt_passwd.pw_gid = get_rid (group_token.PrimaryGroup);
 	      dlength = sizeof (domain);
 	      if (lookup_account_sid (NULL, group_token.PrimaryGroup,
-				      gname, &glength, NULL, &dlength,
+				      gname, &glength, domain, &dlength,
 				      &user_type))
 		strcpy (dflt_group_name, gname);
 	    }

-- 
KOBAYASHI Yasuhiro <kobayays@gmail.com>

Re: Build error with _UNICODE on w32.

[Jason Rumney]

KOBAYASHI Yasuhiro wrote:
> I was trying to maintain the IME-patch on w32. I need it to
> input the unicode character.
>   

Unicode input from IMEs is already supported in 23.0.60. You probably 
need to rewrite your patch to take this into account.

Re: Build error with _UNICODE on w32.

[Eli Zaretskii]

> From: KOBAYASHI Yasuhiro <kobayays@gmail.com>
> Cc: emacs-pretest-bug@gnu.org
> Date: Tue, 20 May 2008 15:44:10 +0900
> 
> The following is ad hoc one.
> 
> w32.c:
> @@ -773,7 +773,7 @@
>  	      dflt_passwd.pw_gid = get_rid (group_token.PrimaryGroup);
>  	      dlength = sizeof (domain);
>  	      if (lookup_account_sid (NULL, group_token.PrimaryGroup,
> -				      gname, &glength, NULL, &dlength,
> +				      gname, &glength, domain, &dlength,
>  				      &user_type))
>  		strcpy (dflt_group_name, gname);
>  	    }

I don't see how this alone could have settled everything.  For
example, this very call to lookup_account_sid should pass wchar_t *
pointers instead of char *, because under UNICODE we call the
wide-character version of the function that returns strings encoded in
UTF-16.  So this call (as well as many others) will simply not work
correctly under UNICODE.

bug#265: Build error with _UNICODE on w32.

[Glenn Morris]

Thanks for the report. Your report was against an old version of Emacs,
and the issue you describe could not be reproduced. The latest version
is 23.3. Please open new reports for any issues you have with the new
version.

bug#265: Build error with _UNICODE on w32.

[Eli Zaretskii]

> From: Glenn Morris <rgm@gnu.org>
> Date: Thu, 10 Mar 2011 01:59:19 -0500
> Cc: 
> 
> Thanks for the report. Your report was against an old version of Emacs,
> and the issue you describe could not be reproduced. The latest version
> is 23.3. Please open new reports for any issues you have with the new
> version.

Actually, I'm sure that this can be reproduced at will.  Building a
Windows port with -D_UNICODE is simply not supported.  I'm not
surprised that the resulting binary crashes.

bug#265: Build error with _UNICODE on w32.

[Glenn Morris]

reopen 265
stop

Eli Zaretskii wrote:

> Actually, I'm sure that this can be reproduced at will.  Building a
> Windows port with -D_UNICODE is simply not supported.  I'm not
> surprised that the resulting binary crashes.

Oh, OK. I'm glad someone actually explained this.

bug#265: Build error with _UNICODE on w32.

[Glenn Morris]

I don't see a need to keep open this particular report, which was marked
"wontfix" some time ago.