Last evening's 0.17 infloops in find-charset-string

Last evening's 0.17 infloops in find-charset-string

[Karl Kleinpaste]

XEmacs 21.4.10 (mule) w/Oort 0.17 updated immediately after 0.16
release was announced.

Very long articles will no longer display at all for me, as
find-charset-string is reproducibly getting stuck.  This wasn't
happening with Oort until I updated last evening.  (I hadn't
previously done "cvs update" in perhaps 10 days.)

Interestingly, the 0.16 release announcement itself is one such
article which will not display due to find-charset-string getting
stuck.  I let XEmacs sit for about 3 minutes, waiting for this article
to display; it never did, and I caught the stack trace with
debug-on-quit.

Signaling: (quit)
  find-charset-string("Mainly mostly misc mush.\n\nGnus 5.10 is probably ...\n\n")
  mm-decode-coding-region-safely(1910 37207 iso-8859-1)
  mm-decode-body("iso-8859-1" 8bit "text/plain" nil)
  article-decode-charset()
  run-hooks(article-decode-charset)
  gnus-request-article-this-buffer(52046 "list.ding")
  gnus-article-prepare(52046 nil)
  gnus-summary-display-article(52046 nil)
  gnus-summary-select-article(nil force)
  gnus-summary-show-article(nil)
  call-interactively(gnus-summary-show-article)

Re: Last evening's 0.17 infloops in find-charset-string

[Jesper Harder]

Karl Kleinpaste <karl@charcoal.com> writes:

> XEmacs 21.4.10 (mule) w/Oort 0.17 updated immediately after 0.16
> release was announced.
>
> Very long articles will no longer display at all for me, as
> find-charset-string is reproducibly getting stuck.

I think you just didn't wait long enough -- it did display for me
eventually (with XEmacs 21.4 (patch 6) "Common Lisp").

The reason is that the XEmacs version of `find-charset-string' (written
in Lisp) is incredible slower than the Emacs version (in C).  

I've changed the code to use `find-charset-region' instead, which should
be more efficient in XEmacs.

Re: Last evening's 0.17 infloops in find-charset-string

[Karl Kleinpaste]

Jesper Harder <harder@myrealbox.com> writes:
> I think you just didn't wait long enough -- it did display for me
> eventually (with XEmacs 21.4 (patch 6) "Common Lisp").

The pre-release CVS-0.16 I've been running for a few weeks displays
long articles immediately.  The post-0.16-release CVS-0.17 to which I
upgraded last night pegged my CPU for multiple minutes and never did
cough up the article, and I figure 3 minutes is far, far, far, longer
than any person should ever wait for any such thing.

Is the use of `find-charset-*' entirely new in latest Oort?

Re: Last evening's 0.17 infloops in find-charset-string

[Jesper Harder]

Karl Kleinpaste <karl@charcoal.com> writes:

> Jesper Harder <harder@myrealbox.com> writes:
>> I think you just didn't wait long enough -- it did display for me
>> eventually (with XEmacs 21.4 (patch 6) "Common Lisp").
>
> The pre-release CVS-0.16 I've been running for a few weeks displays
> long articles immediately.  The post-0.16-release CVS-0.17 to which I
> upgraded last night pegged my CPU for multiple minutes and never did
> cough up the article, and I figure 3 minutes is far, far, far, longer
> than any person should ever wait for any such thing.

Yes, of course.  That's why I changed it to `find-charset-region'
instead.  Try to update and see if it's better.

I was only explaining that it's probably not a bug as such in XEmacs,
but just a question of an implementation that's significantly slower
than in Emacs.

> Is the use of `find-charset-*' entirely new in latest Oort?

`find-charset-string' is new.