XEmacs 21.4 p16, seemingly endless, cpu-consuming loop in string-match?!

XEmacs 21.4 p16, seemingly endless, cpu-consuming loop in string-match?!

[Adam Sjøgren]

  Hi.


Recently Debian unstable got new XEmacs packages. Since then, I think,
xemacs on my machine often enters a seemingly endless, cpu-consuming
loop when I enter a group or try to read an article.

When I kill the xemacs process (after say 5 minutes of burning cpu on
a P4 2.4 GHz), I always get a backtrace that starts with:

 Lisp backtrace follows:

   string-match("=\\?\\([^][- ()<>@,;:*\\\"/?.=]+\\)\\(?:\\*[^?]+\\)?\\?\\(B\\|Q\\)\\?\\([!->@-~ ]*\\)\\?=" "=?iso-8859-1?Q?Sj=F8gren?=")
   # bind (word)
   rfc2047-parse-and-decode("=?iso-8859-1?Q?Sj=F8gren?=")
   # (unwind-protect ...)

(The actual string is different, sometimes it is the From:, other
times it is the Subject: - it has, so far, always been encoded).

It doesn't happen every time I enter the same group/read the same
article, only sometimes. Which is, of course, rather troubling.

Any ideas?


 XEmacs 21.4 (patch 16) "Corporate Culture" [Lucid]
 (i386-debian-linux, Mule) of Sun Dec 12 2004 on penell

 No Gnus v0.3 (update from cvs yesterday and re-bytecompiled)


  Best regards,

-- 
 "Snurra min jord igen"                                       Adam Sjøgren
                                                         asjo@koldfront.dk

Re: XEmacs 21.4 p16, seemingly endless, cpu-consuming loop in string-match?!

[Katsumi Yamaoka]

>>>>> In <87u0qibbw2.fsf@koldfront.dk> Adam Sjøgren wrote:

> Recently Debian unstable got new XEmacs packages. Since then, I think,
> xemacs on my machine often enters a seemingly endless, cpu-consuming
> loop when I enter a group or try to read an article.

I experienced the same problem by XEmacs of the versions before
and after releasing 21.4.16.  It makes it hard to test Gnus and
some ELisp packages with it.

> When I kill the xemacs process (after say 5 minutes of burning cpu on
> a P4 2.4 GHz), I always get a backtrace that starts with:

>  Lisp backtrace follows:

>    string-match("=\\?\\([^][- ()<>@,;:*\\\"/?.=]+\\)\\(?:\\*[^?]+\\)?\\?\\(B\\|Q\\)\\?\\([!->@-~ ]*\\)\\?=" "=?iso-8859-1?Q?Sj=F8gren?=")
>    # bind (word)
>    rfc2047-parse-and-decode("=?iso-8859-1?Q?Sj=F8gren?=")
>    # (unwind-protect ...)

> (The actual string is different, sometimes it is the From:, other
> times it is the Subject: - it has, so far, always been encoded).

> It doesn't happen every time I enter the same group/read the same
> article, only sometimes. Which is, of course, rather troubling.

I'm not sure whether it is related to neither the regexp
matching nor the rfc2047 decoder.  I recommend you to go to the
xemacs-beta list if you still want to continue using XEmacs.  I
lost interest to that long ago.  Sorry.