emacs 22.1.50 death on 64-bit Windows

emacs 22.1.50 death on 64-bit Windows

[Kevin Layer]

Start up the 32-bit version of emacs (I know of no 64-bit version)
with -q and eval this in *scratch*:

   (start-process "foo" "*foo*" "c:/windows/notepad.exe")

and your emacs will disappear after a couple of seconds.

The exact version of Windows: Server 2003 R2, Standard x64 Edition, SP
2.

I have seen it on XP Pro, too.

I have seen it on emacs 21.4, too.

Ideas for debugging?

Re: emacs 22.1.50 death on 64-bit Windows

[Eli Zaretskii]

> Date: Mon, 21 May 2007 16:11:01 -0700
> From: Kevin Layer <layer@franz.com>
> 
> Start up the 32-bit version of emacs (I know of no 64-bit version)
> with -q and eval this in *scratch*:
> 
>    (start-process "foo" "*foo*" "c:/windows/notepad.exe")
> 
> and your emacs will disappear after a couple of seconds.

Problems in CVS versions of Emacs are better reported to
emacs-devel@gnu.org, not here.  This list is for released versions of
Emacs.

> Ideas for debugging?

First, can you try the latest pretest of Emacs 22.1?  You can find it
here:

  ftp://alpha.gnu.org/gnu/emacs/pretest/emacs-22.0.990.tar.gz

It is more important now to fix bugs in the pretest because we expect
to release v22.1 very soon.  By contrast, v22.1.50 is the development
version that will not be released for at least a few months.

If the above pretest also crashes on 64-bit Windows, the best way to
debug this is to run Emacs under a debugger (GDB if your Emacs was
compiled with GCC or Visual Studio if it was compiled by MSVC).

Re: emacs 22.1.50 death on 64-bit Windows

[Eli Zaretskii]

> Date: Mon, 21 May 2007 16:11:01 -0700
> From: Kevin Layer <layer@franz.com>
> 
>    (start-process "foo" "*foo*" "c:/windows/notepad.exe")

Btw, is this command accurate?  Notepad lives in c:/windows/system32,
not in c:/windows.  Is the above exactly what you typed?

Re: emacs 22.1.50 death on 64-bit Windows

[Kevin Layer]

Eli Zaretskii <eliz@gnu.org> wrote:

>> > Date: Mon, 21 May 2007 16:11:01 -0700
>> > From: Kevin Layer <layer@franz.com>
>> > 
>> >    (start-process "foo" "*foo*" "c:/windows/notepad.exe")
>> 
>> Btw, is this command accurate?  Notepad lives in c:/windows/system32,
>> not in c:/windows.  Is the above exactly what you typed?

Yes.

c:\WINDOWS>dir /s notepad.exe
 Volume in drive C is HOBART_C
 Volume Serial Number is 3038-22AF

 Directory of c:\WINDOWS

08/04/2004  05:00 AM            69,120 NOTEPAD.EXE
               1 File(s)         69,120 bytes

 Directory of c:\WINDOWS\system32

08/04/2004  05:00 AM            69,120 notepad.exe
               1 File(s)         69,120 bytes

 Directory of c:\WINDOWS\system32\dllcache

08/04/2004  05:00 AM            69,120 notepad.exe
               1 File(s)         69,120 bytes

     Total Files Listed:
               3 File(s)        207,360 bytes
               0 Dir(s)  10,087,305,216 bytes free

c:\WINDOWS>

Re: emacs 22.1.50 death on 64-bit Windows

[Kevin Layer]

Eli Zaretskii <eliz@gnu.org> wrote:

>> > Date: Mon, 21 May 2007 16:11:01 -0700
>> > From: Kevin Layer <layer@franz.com>
>> > 
>> > Start up the 32-bit version of emacs (I know of no 64-bit version)
>> > with -q and eval this in *scratch*:
>> > 
>> >    (start-process "foo" "*foo*" "c:/windows/notepad.exe")
>> > 
>> > and your emacs will disappear after a couple of seconds.
>> 
>> Problems in CVS versions of Emacs are better reported to
>> emacs-devel@gnu.org, not here.  This list is for released versions of
>> Emacs.

It happens in currently release versions, too.

>> 
>> > Ideas for debugging?
>> 
>> First, can you try the latest pretest of Emacs 22.1?  You can find it
>> here:
>> 
>>   ftp://alpha.gnu.org/gnu/emacs/pretest/emacs-22.0.990.tar.gz

I will download and test this today.

I assume this is a source and not binary distribution.  Is is supposed
to build on 64-bit Windows, or do I do the "build on 32-bit and run on
64-bit" thing?

I also wonder how the above differs from what's in CVS.  I checked my
version out of CVS yesterday.

>> 
>> It is more important now to fix bugs in the pretest because we expect
>> to release v22.1 very soon.  By contrast, v22.1.50 is the development
>> version that will not be released for at least a few months.
>> 
>> If the above pretest also crashes on 64-bit Windows, the best way to
>> debug this is to run Emacs under a debugger (GDB if your Emacs was
>> compiled with GCC or Visual Studio if it was compiled by MSVC).

Yes, I know that's the best way, but it does require knowledge of
emacs internals that I do not possess.

Re: emacs 22.1.50 death on 64-bit Windows

[Jason Rumney]

Kevin Layer wrote:
>
> I assume this is a source and not binary distribution.  Is is supposed
> to build on 64-bit Windows, or do I do the "build on 32-bit and run on
> 64-bit" thing?
>
>   

Emacs does not build as a 64 bit executable on Windows. The request for
help in ftp.gnu.org:/gnu/emacs/windows/README to make Emacs a native 64
bit application on AMD64 and Itanium Windows platforms still stands
(though the later seems to be dying off so it is probably best to
concentrate on the former).

Re: emacs 22.1.50 death on 64-bit Windows

[Eli Zaretskii]

> cc: bug-gnu-emacs@gnu.org
> Comments: In-reply-to Eli Zaretskii <eliz@gnu.org>
>    message dated "Tue, 22 May 2007 09:44:52 +0300."
> Date: Tue, 22 May 2007 07:39:27 -0700
> From: Kevin Layer <layer@franz.com>
> 
> >> First, can you try the latest pretest of Emacs 22.1?  You can find it
> >> here:
> >> 
> >>   ftp://alpha.gnu.org/gnu/emacs/pretest/emacs-22.0.990.tar.gz
> 
> I will download and test this today.
> 
> I assume this is a source and not binary distribution.  Is is supposed
> to build on 64-bit Windows, or do I do the "build on 32-bit and run on
> 64-bit" thing?

The latter.  If you use the MinGW port of GCC to build, you don't have
an alternative, anyway, as there's no 64-bit of that yet.

> I also wonder how the above differs from what's in CVS.  I checked my
> version out of CVS yesterday.

You checked out from the trunk, while the pretests are delivered from
a release branch.  The differences are not very large, so if you
prefer to debug this in the CVS version, go ahead with that.

> >> If the above pretest also crashes on 64-bit Windows, the best way to
> >> debug this is to run Emacs under a debugger (GDB if your Emacs was
> >> compiled with GCC or Visual Studio if it was compiled by MSVC).
> 
> Yes, I know that's the best way, but it does require knowledge of
> emacs internals that I do not possess.

But others do, and if you produce information from the debug session,
they will guide you through the process.

But please, if you decide to give it a go, report the results on
emacs-devel@gnu.org, not here.

Thanks in advance.