Re: Help: build emacs in cygwin (fwd)

Re: Help: build emacs in cygwin (fwd)

[Angelo Graziosi]

Regarding the 'MAKE' questions, I would suggest to try the Cygwin patched
version of make that accepts PATH in DOS style:

  http://cygwin.com/ml/cygwin/2006-11/msg00490.html

If I remeber correcly those patches were accepted upstream.


Cheers,

   Angelo.

Re: Help: build emacs in cygwin (fwd)

[Eli Zaretskii]

> Date: Thu, 8 Mar 2007 12:27:03 +0100 (MET)
> From: Angelo Graziosi <Angelo.Graziosi@roma1.infn.it>
> cc: Eli Zaretskii <eliz@gnu.org>
> 
>   http://cygwin.com/ml/cygwin/2006-11/msg00490.html
> 
> If I remeber correcly those patches were accepted upstream.

Yes, they were.

Re: Help: build emacs in cygwin

[Angelo Graziosi]

Eli Zaretskii wrote:

> Strange, that's the version that caused lots of trouble to Angelo
> Graziosi.  Also, he reported problems with debug info generated by
> 3.4.4-2 (can you debug your binary?).

I confirm. 

I have done a build with GCC-3.4.4 (that now is 3.4.4-3, not -2) using the
configure options described here
http://lists.gnu.org/archive/html/emacs-devel/2007-03/msg00722.html
(it differs from mine only because it does not uses XAW3D) but it segment
faults in a few seconds/minutes!

Every time I use GCC-3.4.4 the result is unstable. Instead GCC-4.0.3 gives
results very stable.


Cheers,

   Angelo. 

Re: Help: build emacs in cygwin

[Harald Maier]

Angelo Graziosi <Angelo.Graziosi@roma1.infn.it> writes:

> Eli Zaretskii wrote:
>
>> Strange, that's the version that caused lots of trouble to Angelo
>> Graziosi.  Also, he reported problems with debug info generated by
>> 3.4.4-2 (can you debug your binary?).
>
> I confirm. 
>
> I have done a build with GCC-3.4.4 (that now is 3.4.4-3, not -2) using the
> configure options described here
> http://lists.gnu.org/archive/html/emacs-devel/2007-03/msg00722.html
> (It differs from mine only because it does not uses XAW3D) but it segment
> faults in a few seconds/minutes!
>
> Every time I use GCC-3.4.4 the result is unstable. Instead GCC-4.0.3 gives
> results very stable.

Me too. An Cygwin-Emacs compiled with gcc-3.4.4 is unusable!
Especially, I saw lots of crashes by subprocess handling such as ssh
and with SQLi buffers.

Harald

Re: Help: build emacs in cygwin

[Angelo Graziosi]

Eli Zaretskii wrote:

>> From: "djh" <address@hidden>
>> Date: Tue, 13 Mar 2007 14:51:48 +0900
>> 
>> No. I get the below results.  Same as Angelo, but I think Angelo was
>> trying 
>> to build a native w32 version and had pathname converstion problems.

> No, he was building a Cygwin binary, not a native w32 binary.

Obviously.

I never tried to build a native 32 binary.

I refer always to a pure Cygwin build.


Cheers,

   Angelo.

Re: Help: build emacs in cygwin

[djh]

I did not have the problems that Angelo had. (I do have crash problems, but almost exclusively when using ddskk, a skk based Japanese input method.). 

I believe Angelo also had symbol problems, that is a lack of symbols for debugging.

I believe the problem is not the compiler, but the cygwin x-window implementation.
If if the compiler, then it has been fixed (my 3.4.4 shows an installed date of December 18, 2006).

I mainly have problems (crashes) when using a Japanese input method.  
Which leads me more to believe that the problem is not the compiler but the cygwin x-window implementation. 

> Me too. An Cygwin-Emacs compiled with gcc-3.4.4 is unusable!
> Especially, I saw lots of crashes by subprocess handling such as ssh
> and with SQLi buffers.
> 
> Harald

I just downloaded and built the newest gcc, gcc (GCC) 4.1.3 20070312 (prerelease)
and compiled emacs with it successfully.

I still get:
$ gdb /bin/emacs.exe
GNU gdb 6.5.50.20060706-cvs (cygwin-special)
Copyright (C) 2006 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i686-pc-cygwin"...
(gdb) r
Starting program: /bin/emacs.exe
Loaded symbols for /cygdrive/c/WINDOWS/system32/ntdll.dll
Loaded symbols for /cygdrive/c/WINDOWS/system32/kernel32.dll
Loaded symbols for /usr/bin/cygncurses-8.dll
Loaded symbols for /usr/bin/cygwin1.dll
Loaded symbols for /cygdrive/c/WINDOWS/system32/advapi32.dll
Loaded symbols for /cygdrive/c/WINDOWS/system32/rpcrt4.dll
Loaded symbols for /usr/bin/cygjpeg-62.dll
Loaded symbols for /usr/bin/cygpng12.dll
Loaded symbols for /usr/bin/cygz.dll
Loaded symbols for /usr/bin/cygtiff-5.dll
Loaded symbols for /usr/bin/cygungif-4.dll
Loaded symbols for /usr/X11R6/bin/cygX11-6.dll
Loaded symbols for /usr/X11R6/bin/cygICE-6.dll
Loaded symbols for /usr/X11R6/bin/cygSM-6.dll
Loaded symbols for /usr/X11R6/bin/cygXaw-8.dll
Loaded symbols for /usr/X11R6/bin/cygXext-6.dll
Loaded symbols for /usr/X11R6/bin/cygXmu-6.dll
Loaded symbols for /usr/X11R6/bin/cygXt-6.dll
Loaded symbols for /usr/X11R6/bin/cygXp-6.dll
Loaded symbols for /usr/X11R6/bin/cygXau-6.dll
Loaded symbols for /usr/X11R6/bin/cygXpm-4.dll
emacs: standard input is not a tty

Program exited with code 01.

-----

Which leads me to believe it is not a compiler problem.

Darel

Re: Help: build emacs in cygwin

[Jason Rumney]

djh wrote:
> I mainly have problems (crashes) when using a Japanese input method.  
> Which leads me more to believe that the problem is not the compiler but the cygwin x-window implementation. 
>   

I think there is a bug in the interaction between skk and some versions 
of X and/or certain window managers. I've seen lockups and crashes on 
GNU/Linux with certain combinations not involving Emacs or Cygwin.

Re: emacs snapshot tarball test build

[Angelo Graziosi]

djh wrote:

> The only problem is shown below (and maybe it is not a real problem).
>
> --------
> make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/leim'
> cd lib-src; make maybe-blessmail  \
>          MAKE='make' archlibdir='/libexec/emacs/22.0.95/i686-pc-cygwin'
> make[1]: Entering directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> ../src/emacs -batch --no-site-file --multibyte -l 
> /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> make: *** [blessmail] Error 2
> ------


This often happens boostraping Emacs on Cygwin.

Workaround: try to build in another nested directory.

Generally, a build with the above problem gives problems!


Cheers,

   Angelo.

Re: emacs snapshot tarball test build

[djh]

Thanks Angelo.  But, one shouldn't have to build in a specific directory, or guess about a directory structure, and this should be an easy thing to fix. 

If it is a semi-hard coded (if should a thing code be) reference or assumption in a Makefile then it should be corrected in my opinion.

It would be good to know why "make" aborted on blessmail.el.  It was reported as
"Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)" below.

> djh wrote:
> > The only problem is shown below (and maybe it is not a real problem).
> > --------
> > make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/leim'
> > cd lib-src; make maybe-blessmail  \
> >          MAKE='make' archlibdir='/libexec/emacs/22.0.95/i686-pc-cygwin'
> > make[1]: Entering directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> > ../src/emacs -batch --no-site-file --multibyte -l 
> > /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> > Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> > make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> > make: *** [blessmail] Error 2
> > ------
> 
> 
> This often happens boostraping Emacs on Cygwin.
> Workaround: try to build in another nested directory.
> Generally, a build with the above problem gives problems!
> 
> Cheers,
>    Angelo.

Regards,
   Darel Henman

Re: emacs snapshot tarball test build

[Angelo Graziosi]

On Fri, 23 Mar 2007, djh wrote:

> 
> Thanks Angelo.  But, one shouldn't have to build in a specific directory, or guess about a directory structure, and this should be an easy thing to fix. 
> 
> If it is a semi-hard coded (if should a thing code be) reference or assumption in a Makefile then it should be corrected in my opinion.
> 
> It would be good to know why "make" aborted on blessmail.el.  It was reported as


> "Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)" below.


Boostrapping Emacs on Cygwin, one often has "Fatal error
(6)make[1]:..." not only at the end, when installing, with "blessmail",
but also when it starts to compile "el" files or after it has rebuilt the
final version of "emacs.exe".

Theste errors happen at random, without an evident logic.

Usually, changing the build directory (increasing the nested level) solve
the thing (same Emacs CVS source).

So correcting the Makefile left some doubts (but I can err).

Obviously, the bootstrap of the same CVS code on GNU/Linux Kubuntu does
not give any problem.


Cheers,

   Angelo.
 







> 
> > djh wrote:
> > > The only problem is shown below (and maybe it is not a real problem).
> > > --------
> > > make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/leim'
> > > cd lib-src; make maybe-blessmail  \
> > >          MAKE='make' archlibdir='/libexec/emacs/22.0.95/i686-pc-cygwin'
> > > make[1]: Entering directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> > > ../src/emacs -batch --no-site-file --multibyte -l 
> > > /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> > > Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> > > make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> > > make: *** [blessmail] Error 2
> > > ------
> > 
> > 
> > This often happens boostraping Emacs on Cygwin.
> > Workaround: try to build in another nested directory.
> > Generally, a build with the above problem gives problems!
> > 
> > Cheers,
> >    Angelo.
> 
> Regards,
>    Darel Henman
> 

Re: emacs snapshot tarball test build

[Eli Zaretskii]

> From: "djh" <henman@it.to-be.co.jp>
> Date: Fri, 23 Mar 2007 10:34:19 +0900
> Cc: djh <henman@it.to-be.co.jp>, emacs-devel@gnu.org
> 
> It would be good to know why "make" aborted on blessmail.el.  It was reported as
> "Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)" below.

It's not "make" who aborted, it's Emacs.  "make" just printed a
human-readable description of why Emacs died.

Re: Pretest 22.0.97

[Angelo Graziosi]

Cygwin binaries here:

   http://www.webalice.it/angelo.graziosi/Emacs.html


Cheers,

  Angelo.

Re: Build failure on Cygwin

[Angelo Graziosi]

Eric Hanchrow wrote:

> I simply did "cvs co ...", "./configure", "make bootstrap". 
>
> ...
> Fatal error (6)/bin/sh: line 2:   220 Aborted....
>

This often happens in bootstrapping under Cygwin.

Solution: build in a separate directory and/or build from an
"xterm" console (NOT the dos-like console)...

...or use prebuild (22.0.97-pretest) Cygwin binaries from
http://www.webalice.it/angelo.graziosi/Emacs.html.

To build use GCC-4.0.3 or 4.0.4 which seem to give very stable binaries.



Cheers,

   Angelo.

RE: merging Unicode branch and availability of Windows binaries

[Angelo Graziosi]

Drew Adams wrote:

> I asked about Windows binaries. Where will one be able to find a Windows
> binary of Emacs 22?

You will find the Cygwin binaries (which are in any case Windows
binaries!) here:


  http://www.webalice.it/angelo.graziosi/Emacs.html


For the moment, there you will find only Emacs pretest 22.0.99 waiting for
the release 22!



Cheers,

   Angelo.

RE: merging Unicode branch and availability of Windows binaries

[Drew Adams]

> > Where will one be able to find a Windows binary of Emacs 22?
>
> You will find the Cygwin binaries (which are in any case Windows
> binaries!) here: http://www.webalice.it/angelo.graziosi/Emacs.html

Thanks, Angelo. Actually, I was looking for non-Cygwin (though I didn't
state that), but this is also good to know.

Re: emacs snapshot tarball test build

[Richard Stallman]

    This often happens boostraping Emacs on Cygwin.

    Workaround: try to build in another nested directory.

Is this documented in the installation instructions?
In etc/PROBLEMS?

If not, would someone like to add that info?

Re: emacs snapshot tarball test build

[Eli Zaretskii]

> From: Richard Stallman <rms@gnu.org>
> Date: Fri, 23 Mar 2007 00:37:11 -0400
> Cc: henman@it.to-be.co.jp, emacs-devel@gnu.org
> 
>     This often happens boostraping Emacs on Cygwin.
> 
>     Workaround: try to build in another nested directory.
> 
> Is this documented in the installation instructions?
> In etc/PROBLEMS?
> 
> If not, would someone like to add that info?

These are not workarounds that I'd recommend adding to PROBLEMS.

In general, if the problem only happens during bootstrap, and doesn't
have a good workarounds, I'd say let's ignore it.  Most Emacs users
will never bootstrap.

emacs snapshot tarball test build

[djh]

I was asked to try out building a snapshot tarball on cygwin and have just finished.

Platform: 
   gcc (GCC) 4.1.3 20070312 (prerelease) 
   CYGWIN_NT-5.1 1.5.24(0.156/4/2) 2007-01-31 10:57

Results:
    configured and built fine.

Install did not complete successfully, but did install a workable newly built emacs.

The only problem is shown below (and maybe it is not a real problem).

--------
make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/leim'
cd lib-src; make maybe-blessmail  \
          MAKE='make' archlibdir='/libexec/emacs/22.0.95/i686-pc-cygwin'
make[1]: Entering directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
../src/emacs -batch --no-site-file --multibyte -l /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
make: *** [blessmail] Error 2
------

However the newly installed emacs seems to work and I am writing this letter with it and mh-e.

What comes after "blessmail" in the Make sequence I am not familiar with, but suspect that it is near the end of the install procedure if I recall correctly.

Regards,
 Darel Henman

Re: emacs snapshot tarball test build

[Eli Zaretskii]

> From: "djh" <henman@it.to-be.co.jp>
> Date: Thu, 22 Mar 2007 15:10:36 +0900
> 
> I was asked to try out building a snapshot tarball on cygwin and have just finished.

Thanks.

> Install did not complete successfully, but did install a workable newly built emacs.
> 
> The only problem is shown below (and maybe it is not a real problem).
> 
> --------
> make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/leim'
> cd lib-src; make maybe-blessmail  \
>           MAKE='make' archlibdir='/libexec/emacs/22.0.95/i686-pc-cygwin'
> make[1]: Entering directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> ../src/emacs -batch --no-site-file --multibyte -l /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> make[1]: Leaving directory `/usr/src/emacs-tmp/emacs-22.0.95/lib-src'
> make: *** [blessmail] Error 2

To me, this looks as an evidence that the Emacs binary you built is
unreliable.  So I think we should not yet add GCC 4.1.3 to the the 3
safe versions of Cygwin GCC mentioned in PROBLEMS.

Thanks for working on this.

Re: emacs snapshot tarball test build

[djh]

> > The only problem is shown below (and maybe it is not a real problem).
> > --------
> > ../src/emacs -batch --no-site-file --multibyte -l /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> > Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> 
> To me, this looks as an evidence that the Emacs binary you built is
> unreliable.  So I think we should not yet add GCC 4.1.3 to the the 3
> safe versions of Cygwin GCC mentioned in PROBLEMS.

GCC 4.1.3 (actually gcc-core-4.1-20070312.tar.bz2 and friends) does provide for a 
robust build from cvs though (or did on March 13th anyway.)

Darel

Re: emacs snapshot tarball test build

[Eli Zaretskii]

> Cc: "djh" <henman@it.to-be.co.jp>, <emacs-devel@gnu.org>
> From: "djh" <henman@it.to-be.co.jp>
> Date: Wed, 28 Mar 2007 16:32:49 +0900
> 
> > > The only problem is shown below (and maybe it is not a real problem).
> > > --------
> > > ../src/emacs -batch --no-site-file --multibyte -l /usr/src/emacs-tmp/emacs-22.0.95/lib-src/../lisp/mail/blessmail.el
> > > Fatal error (6)make[1]: *** [blessmail] Aborted (core dumped)
> > 
> > To me, this looks as an evidence that the Emacs binary you built is
> > unreliable.  So I think we should not yet add GCC 4.1.3 to the the 3
> > safe versions of Cygwin GCC mentioned in PROBLEMS.
> 
> GCC 4.1.3 (actually gcc-core-4.1-20070312.tar.bz2 and friends) does provide for a 
> robust build from cvs though (or did on March 13th anyway.)

How can you call it ``robust'' when it crashes while loading
blessmail?

Re: emacs snapshot tarball test build

[djh]

The cvs build with 4.1.20070312 did not crash when loading blessmail, and 
it hasn't yet crashed when using skk input method which gcc (GCC) 3.4.4 (cygming special, gdc 0.12, using dmd 0.125) did frequently.

> > 
> > GCC 4.1.3 (actually gcc-core-4.1-20070312.tar.bz2 and friends) does provide for a 
> > robust build from cvs though (or did on March 13th anyway.)
> 
> How can you call it ``robust'' when it crashes while loading
> blessmail?

So my question is why does it build cvs emacs well and not the bz2'd version.
To be accurate I should get a tarball the exact same date as the cvs version.

Darel Henman