heisenbugs 1 joakim 0

heisenbugs 1 joakim 0

[joakim]

Hello,

Currently Emacs heisenbugs are having a field day in my computer. I'd
like to request some assistance in evening the score if possible.

These are heisenbugs and furthermore in my own xwidget branch, so if
all this sounds like useless whining, please just disregard. A guy needs
to went now and then.

- The recent bidi change seemed to introduce some visual artifacts
  sometimes during scrolling. Sometimes the screen gets scrolled in
  small rectangular regions but not all areas. It could be my branch
  code but I dont think so, because my code is the bestest. It could
  also be GTK3 + Bidi. This heisenbug is endurable because I can make
  the display right by requesting a redraw.

- Gnus usualy hangs for me with a flaky network but its much worse
  recently with the gnutls stuff. Sometimes I get a message like:

"gnutls.c [0] (Emacs) fatal error: The specified reason has been
invalidated for some reason."

- I run Emacs in gdb all the time but it doesn't really help
  recently. Emacs hangs so hard gdb can't interrupt. Sometimes I can
  at least get a backtrace by sending a SIGHUP signal. Usualy the hang
  is in accept_process_output initiated from Gnus.

- bzr currently hangs, at the savannah end(Okay, I just like to whine
  a bit)

- Hey, Martins new window code at least doesn't give me troubles!

- Since Emacs now crashes/hangs many times a day, desktop mode helps
  me a lot. Can I make it save the state more often?

-- 
Joakim Verona

Re: heisenbugs 1 joakim 0

[Lars Magne Ingebrigtsen]

joakim@verona.se writes:

> - Gnus usualy hangs for me with a flaky network but its much worse
>   recently with the gnutls stuff. Sometimes I get a message like:
>
> "gnutls.c [0] (Emacs) fatal error: The specified reason has been
> invalidated for some reason."

I haven't seen that message, but my network connections aren't flaky... 

> - I run Emacs in gdb all the time but it doesn't really help
>   recently. Emacs hangs so hard gdb can't interrupt. Sometimes I can
>   at least get a backtrace by sending a SIGHUP signal. Usualy the hang
>   is in accept_process_output initiated from Gnus.

Does strace say what it's doing?

> - bzr currently hangs, at the savannah end(Okay, I just like to whine
>   a bit)

While we're whining -- there are still no bzr diff messages being
generated, which is awkward.

Oh, and I'm chasing down a heisenbug, too -- sometimes (about one in
fifty times), Emacs crashes while displaying animated GIFs.  This
started about a week or so ago.  I still haven't been able to catch the
bug happening while running under gdb.

-- 
(domestic pets only, the antidote for overdose, milk.)
  bloggy blog http://lars.ingebrigtsen.no/

Re: heisenbugs 1 joakim 0

[Davis Herring]

> - Since Emacs now crashes/hangs many times a day, desktop mode helps
>   me a lot. Can I make it save the state more often?

(run-with-timer 0 60 'desktop-save-in-desktop-dir)

Since Emacs isn't normally this buggy, I don't know that we should add
this to desktop.el, but we certainly could.

Davis

-- 
This product is sold by volume, not by mass.  If it appears too dense or
too sparse, it is because mass-energy conversion has occurred during
shipping.

Re: heisenbugs 1 joakim 0

[Antoine Levitt]

02/08/11 15:39, joakim@verona.se
> Hello,
>
> Currently Emacs heisenbugs are having a field day in my computer. I'd
> like to request some assistance in evening the score if possible.
>
> These are heisenbugs and furthermore in my own xwidget branch, so if
> all this sounds like useless whining, please just disregard. A guy needs
> to went now and then.
>
> - The recent bidi change seemed to introduce some visual artifacts
>   sometimes during scrolling. Sometimes the screen gets scrolled in
>   small rectangular regions but not all areas. It could be my branch
>   code but I dont think so, because my code is the bestest. It could
>   also be GTK3 + Bidi. This heisenbug is endurable because I can make
>   the display right by requesting a redraw.
>
> - Gnus usualy hangs for me with a flaky network but its much worse
>   recently with the gnutls stuff. Sometimes I get a message like:
>
> "gnutls.c [0] (Emacs) fatal error: The specified reason has been
> invalidated for some reason."

I do sometimes get error messages like these (actually, not like this,
they're "size zero" errors). I also get random hangs on
gnus-group-get-new-news. Strange thing is that my gnus only connects to
local servers (dovecot & leafnode) ... maybe it's some error at the
other end. I've added some code to get backtraces, I'll see if I can get
something usable and report it.

>
> - I run Emacs in gdb all the time but it doesn't really help
>   recently. Emacs hangs so hard gdb can't interrupt. Sometimes I can
>   at least get a backtrace by sending a SIGHUP signal. Usualy the hang
>   is in accept_process_output initiated from Gnus.
>
> - bzr currently hangs, at the savannah end(Okay, I just like to whine
>   a bit)

There's a thread on this, the savannah admins have been notified so
presumably something will be done in the near future.

>
> - Hey, Martins new window code at least doesn't give me troubles!
>
> - Since Emacs now crashes/hangs many times a day, desktop mode helps
>   me a lot. Can I make it save the state more often?

I've got this in my .emacs

(run-with-timer (* 10 60) (* 10 60) (lambda () (flet ((message (&rest args) nil))
						   (desktop-save-in-desktop-dir))))
                                                   
(the flet hack is there to suppress any output)

Re: heisenbugs 1 joakim 0

[Andreas Schwab]

joakim@verona.se writes:

> Currently Emacs heisenbugs are having a field day in my computer. I'd
> like to request some assistance in evening the score if possible.

Start with the warnings.

Andreas.

-- 
Andreas Schwab, schwab@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: heisenbugs 1 joakim 0

[Eli Zaretskii]

> From: joakim@verona.se
> Date: Tue, 02 Aug 2011 15:39:07 +0200
> 
> - The recent bidi change seemed to introduce some visual artifacts
>   sometimes during scrolling.

Do these artifacts disappear if you set bidi-display-reordering to a
nil value?  If they don't disappear, then the probability that they
are caused by bidi are low, because setting that variable to nil
disables most of the bidi code.  (The probability is low to begin
with, because bidi changes didn't touch the display back-ends, where
pixel drawing happens.)

bzr diff messages (was: heisenbugs 1 joakim 0)

[Reiner Steib]

On Tue, Aug 02 2011, Lars Magne Ingebrigtsen wrote:

> While we're whining -- there are still no bzr diff messages being
> generated, which is awkward.

At least there is gwene.org.gnu.savannah.bzr.lh.emacs.trunk ...

Bye, Reiner.
-- 
       ,,,
      (o o)
---ooO-(_)-Ooo---  |  PGP key available  |  http://rsteib.home.pages.de/

Re: heisenbugs 1 joakim 0

[joakim]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: joakim@verona.se
>> Date: Tue, 02 Aug 2011 15:39:07 +0200
>> 
>> - The recent bidi change seemed to introduce some visual artifacts
>>   sometimes during scrolling.
>
> Do these artifacts disappear if you set bidi-display-reordering to a
> nil value?  If they don't disappear, then the probability that they
> are caused by bidi are low, because setting that variable to nil
> disables most of the bidi code.  (The probability is low to begin
> with, because bidi changes didn't touch the display back-ends, where
> pixel drawing happens.)

This had nothing to do with bidi. I started playing a game with my son
which required some experimental direct rendering settings. This
aparently affected the 3D accelerated WM in Gnome3, so scrolling of
window text got affected. So, sorry for the noise.

-- 
Joakim Verona

Re: heisenbugs 1 joakim 0

[joakim]

"Davis Herring" <herring@lanl.gov> writes:

>> - Since Emacs now crashes/hangs many times a day, desktop mode helps
>>   me a lot. Can I make it save the state more often?
>
> (run-with-timer 0 60 'desktop-save-in-desktop-dir)
>
> Since Emacs isn't normally this buggy, I don't know that we should add
> this to desktop.el, but we certainly could.

I've started to use the new theme facilities for more than just
appearances. my zen-mode.el sets stuff like fullscreen mode, httpd proxy
and such.

Maybe we could make a emacs-developer-theme.el with various settings like these?

>
> Davis

-- 
Joakim Verona