Re: Strange Emacs crash

[Kim F. Storm]

Reiner Steib <address@hidden> writes:

> On Fri, Feb 25 2005, Kim F. Storm wrote:
>
>> Richard Stallman <address@hidden> writes:
>>
>>>     It is a problem of a function called from key-translation-map 
>>>     doing read-event followed by modifying the buffer.
>>>
>>> I looked at the code in read_key_sequence, keyremap_step, and
>>> access_keymap_keyremap.  I don't see anything that interacts specially
>>> with redisplay.  So it is a mystery why this would behave any differently
>>> from calling read-event and modifying the buffer in some other context.
>>>
>>> Can you find any clue to why doing it here is special?
>>
>> Since I have installed a fix for the specific crash (after spending
>> several hours on the problem), I think I have other more urgent
>> crashes to investigate...
>>
>> I suggest we put it on the to-do list for investigation AFTER
>> the release.
>
> JFTR: This problem sounds similar to the problem (Double mode crashes
> emacs in message mode) reported by Niels Olof Bouvin:
> http://thread.gmane.org/gmane.emacs.pretest.bugs/5894

Right.

It calls read-event inside a key-translation-map function.
But that function doesn't update the buffer.

However, I can also reproduce the crash, and it seems that
the following patch fixes the problem:

*** double.el   01 Sep 2003 17:45:11 +0200      1.18
--- double.el   25 Feb 2005 22:27:07 +0100      
***************
*** 116,121 ****
--- 116,122 ----
           ;; End of generated event.  See if he will repeat it...
           (let ((new (double-read-event prompt))
                 (entry (assoc double-last-event double-map)))
+            (force-window-update (selected-window))
             (if (eq new double-last-event)
                 (progn
                   (setq unread-command-events



-- 
Kim F. Storm <address@hidden> http://www.cua.dk