Re: my Emacs is broken

Re: my Emacs is broken

[Richard Stallman]

Karl Zeitler said that this change was necessary to fix a problem for
him on Solaris.  If it causes some other problem, I could remove the
change, but that would bring back the problem he fixed.  It would be
better if we could fix both problems.  However, only Solaris users can
figure out what is really right to do.


Message-Id: <200208172345.g7HNjLH21569@arc.math.ohio-state.edu>
From: Markus Rost <rost@math.ohio-state.edu>
To: rms@gnu.org
Subject: my Emacs is broken
X-Mailer: GNU Emacs 21.2 
X-Home-Page: http://www.math.ohio-state.edu/~rost
X-RAVMilter-Version: 8.3.2(snapshot 20020213) (mathserv)
X-Spam-Status: No, hits=0.0 required=5.0
	tests=none
	version=2.31
X-Spam-Level: 
Content-Length: 472

My Emacs is broken:  It starts an X window, but that is blank and does
not react to any input.  This is in

In GNU Emacs ---current CVS--- (sparc-sun-solaris2.8, X toolkit)

I guess this problem is related with this change:

2002-08-17  Richard M. Stallman  <rms@gnu.org>

	* s/sol2-5.h (BROKEN_SIGIO): Add #undef.

At least, after undoing this change and recompiling, Emacs started out
well.

(In case you ask me for more tests, it may take 2 weeks until I'll
respond.)

Re: my Emacs is broken

[Klaus Zeitler]

Hello Markus,

>>>>> "Richard" == Richard Stallman <rms@gnu.org> writes:
    Richard> 
    Richard> Karl Zeitler said that this change was necessary to fix a problem

I assume he's talking about me (Klaus Zeitler :-)

    Richard> for him on Solaris.  If it causes some other problem, I could
    Richard> remove the change, but that would bring back the problem he
    Richard> fixed.  It would be better if we could fix both problems.
    Richard> However, only Solaris users can figure out what is really right
    Richard> to do.
    Richard> 
    Richard> Message-Id: <200208172345.g7HNjLH21569@arc.math.ohio-state.edu>
    Richard> From: Markus Rost <rost@math.ohio-state.edu>
    Richard> To: rms@gnu.org
    Richard> Subject: my Emacs is broken
    Richard> X-Mailer: GNU Emacs 21.2 
    Richard> X-Home-Page: http://www.math.ohio-state.edu/~rost
    Richard> X-RAVMilter-Version: 8.3.2(snapshot 20020213) (mathserv)
    Richard> X-Spam-Status: No, hits=0.0 required=5.0
    Richard> 	tests=none
    Richard> 	version=2.31
    Richard> X-Spam-Level: 
    Richard> Content-Length: 472
    Richard> 
    Richard> My Emacs is broken:  It starts an X window, but that is blank and
    Richard> does not react to any input.  This is in
    Richard> 
    Richard> In GNU Emacs ---current CVS--- (sparc-sun-solaris2.8, X toolkit)
    Richard> 
    Richard> I guess this problem is related with this change:
    Richard> 
    Richard> 2002-08-17  Richard M. Stallman  <rms@gnu.org>
    Richard> 
    Richard> 	* s/sol2-5.h (BROKEN_SIGIO): Add #undef.
    Richard> 
    Richard> At least, after undoing this change and recompiling, Emacs
    Richard> started out well.
    Richard> 

I've fetched the newest CVS and compiled under Solaris 5.8. Works fine for
me. No problems.

Could you please check/try:

1. you also have the newest sysdep.c, i.e. should contain

#ifndef VMS
#include <sys/file.h>
#endif /* not VMS */

2. check if /usr/include/sys/file.h contains a define for FASYNC

3. run configure and then rebuild


BTW this patch was necessary to make Ctrl-g work in endless loops.
I assume if you're trying something like (while t nil) followed by C-j
then emacs will loop and you can't interrupt.

Klaus


-- 
 ------------------------------------------
|  Klaus Zeitler      Lucent Technologies  |
|  Email:             kzeitler@lucent.com  |
 ------------------------------------------
---
When I was a boy I was told that anybody could become President.
Now I'm beginning to believe it.              -- Clarence Darrow

Re: my Emacs is broken

[Markus Rost]

I can't do any tests for the next weeks.  I'll come back to this then.

Re: my Emacs is broken

[Klaus Zeitler]

>>>>> "Richard" == Richard Stallman <rms@gnu.org> writes:
    Richard> 
    Richard> Karl Zeitler said that this change was necessary to fix a problem
    Richard> for him on Solaris.  If it causes some other problem, I could
    Richard> remove the change, but that would bring back the problem he
    Richard> fixed.  It would be better if we could fix both problems.
    Richard> However, only Solaris users can figure out what is really right
    Richard> to do.
    Richard> 
    Richard> Message-Id: <200208172345.g7HNjLH21569@arc.math.ohio-state.edu>
    Richard> From: Markus Rost <rost@math.ohio-state.edu>
    Richard> To: rms@gnu.org
    Richard> Subject: my Emacs is broken
    Richard> X-Mailer: GNU Emacs 21.2 
    Richard> X-Home-Page: http://www.math.ohio-state.edu/~rost
    Richard> X-RAVMilter-Version: 8.3.2(snapshot 20020213) (mathserv)
    Richard> X-Spam-Status: No, hits=0.0 required=5.0
    Richard> 	tests=none
    Richard> 	version=2.31
    Richard> X-Spam-Level: 
    Richard> Content-Length: 472
    Richard> 
    Richard> My Emacs is broken:  It starts an X window, but that is blank and
    Richard> does not react to any input.  This is in
    Richard> 
    Richard> In GNU Emacs ---current CVS--- (sparc-sun-solaris2.8, X toolkit)
    Richard> 
    Richard> I guess this problem is related with this change:
    Richard> 
    Richard> 2002-08-17  Richard M. Stallman  <rms@gnu.org>
    Richard> 
    Richard> 	* s/sol2-5.h (BROKEN_SIGIO): Add #undef.

When I set the DISPLAY variable to 0:0 I get the same result as Markus
and David Koppelman. If the DISPLAY variable is either set to `X-hostname`:0.0
(that's what I always use) or localhost:0.0 (as long as we're on the X-server),
than everything's fine. David Koppelman just sent me email that it also
works for him when the DISPLAY var is set properly.

Klaus

-- 
 ------------------------------------------
|  Klaus Zeitler      Lucent Technologies  |
|  Email:             kzeitler@lucent.com  |
 ------------------------------------------
---
Real programmers don't comment their code.  It was
hard to write, it should be hard to understand.

Re: my Emacs is broken

[Jan D.]

>     Richard> My Emacs is broken:  It starts an X window, but that is blank and
>     Richard> does not react to any input.  This is in
>     Richard> 
>     Richard> In GNU Emacs ---current CVS--- (sparc-sun-solaris2.8, X toolkit)
>     Richard> 
>     Richard> I guess this problem is related with this change:
>     Richard> 
>     Richard> 2002-08-17  Richard M. Stallman  <rms@gnu.org>
>     Richard> 
>     Richard> 	* s/sol2-5.h (BROKEN_SIGIO): Add #undef.
> 
> When I set the DISPLAY variable to 0:0 I get the same result as Markus
> and David Koppelman. If the DISPLAY variable is either set to `X-hostname`:0.0
> (that's what I always use) or localhost:0.0 (as long as we're on the X-server),
> than everything's fine. David Koppelman just sent me email that it also
> works for him when the DISPLAY var is set properly.

I assume you mean :0.0?  If not, that may be some Sun specific thing.
Anyway, if you set DISPLAY to :0.0 you let the Xlib choose "optimal"
communication mechanism to the X server, whatever that may be.
If you use hostname:0 you usually get TCP/IP over the loopback i/f (i.e.
stream sockets).  Setting DISPLAY to unix:0 gives you unix domain sockets,
which usually is a bit faster than TCP/IP (but not much,
just a few percent on my box).

It is possible that :0.0 on your machine selects some means of communication
that does not work with SIGIO.  Could be shared memory perhaps.  Some
types of X over shared memory requires application support.  Emacs does
not have that.

	Jan D.

Re: my Emacs is broken

[Richard Stallman]

    When I set the DISPLAY variable to 0:0 I get the same result as Markus
    and David Koppelman. If the DISPLAY variable is either set to `X-hostname`:0.0
    (that's what I always use) or localhost:0.0 (as long as we're on the X-server),
    than everything's fine. David Koppelman just sent me email that it also
    works for him when the DISPLAY var is set properly.

That does not bring us any closer to finding out what change
caused the problem to appear.  In general, trying different ways
to invoke a failing program is an ineffective approach to debugging.

Re: my Emacs is broken

[Klaus Zeitler]

>>>>> "Jan" == Jan D <jan.h.d@swipnet.se> writes:

    >> When I set the DISPLAY variable to 0:0 I get the same result as Markus
    >> and David Koppelman. If the DISPLAY variable is either set to
    >> `X-hostname`:0.0 (that's what I always use) or localhost:0.0 (as long
    >> as we're on the X-server), than everything's fine. David Koppelman just
    >> sent me email that it also works for him when the DISPLAY var is set
    >> properly.
    Jan> 
    Jan> I assume you mean :0.0?  If not, that may be some Sun specific thing.

OOPS, yes of course I meant ":0.0".  I've set breakpoints to
XTread_socket, read_avail_input and input_available_signal.
These functions never get called when I set the DISPLAY to ":0.0".
It works when I set the display to "<x-host>:0.0", "localhost:0.0" or
"unix:0.0".

Klaus



-- 
 ------------------------------------------
|  Klaus Zeitler      Lucent Technologies  |
|  Email:             kzeitler@lucent.com  |
 ------------------------------------------
---
The idea is to die young as late as possible. -- Ashley Montagu