Re: Mule redisplay crash

Gunnar Evermann <ge204(a)eng.cam.ac.uk&gt; wrote on 13 Dec 1999: Gunnar offered the following patch, which I am using all the time since then and apply to every new release of XEmacs (21.1 and 21.2): I don't notice any slowdown or any other negative effects. If nothing better can be found, then maybe Gunnars patch could be applied to XEmacs? For my usage pattern, Gunnars patch is essential, without it XEmacs crashes extremely often while using the 'lookup.el' interface to electronic dictionaries. I can reproduce a crash using lookup.el easily by doing the following: ~$ xemacs C-x C-y t e s t (C-x C-y runs `lookup-pattern') C-x 1 (C-x 1 runs `delete-other-windows') C-x C-y t e s t (C-x C-y runs `lookup-pattern') M-x p (M-p runs `lookup-history-previous') => crash. The dictionary I use to reproduce this crash with lookup.el is "Iwanami Shoten Koujien Daiyonban"[1], a Japanese dictionary. To reproduce the crash, a Japanese dictionary is certainly necessary. Gunnar's elisp example code above, which reproduces the crash, uses (insert "が"), which inserts a Japanese character. When using 'lookup.el' wich a Japanese dictionary, similar things like in Gunnar's example code happen all the time. When looking up a few random words in the dictionary, splitting windows a few times, going back and forth in the history of the looked up words, XEmacs will certainly crash soon. That means XEmacs won't usually survive longer than a few minutes if one is using 'lookup.el' with a Japanese dictionary. What do you think? Doesn't it make sense to apply Gunnar's patch? It solves the above problems and doesn't seem to do any harm. Or does anybody know a better solution? Mike Footnotes: [1] 岩波書店　広辞苑（第四版） -- Mike Fabian <mike.fabian(a)gmx.de&gt;