Re: [Ok, with patch] XEmacs 21.5-b20 "cilantro" (+CVS-20050310) s parc-sun-solaris2.8

Adrian> Raymond Toy <raymond.toy(a)ericsson.com&gt; writes: Adrian> Hi Ray! Raymond> I think I should qualify the slowness a bit. It seems Raymond> that gnus is slow. I have some folders in gnus with Raymond> about 1000 unread entries. It seems b20 takes a lot Raymond> longer and gc's quite a bit when getting the summary Raymond> buffer. I think there have been changes to both the regexp code and to font-lock which might affect Gnus. Are you running with error-checking on in Mule? This increases computation by a full factor of N (N = buffer size) to many operations, as with a variable-width representation of characters checking position-computing code may require starting from the beginning.... (setq gc-cons-threshold (* 8 1000 1000)) is probably a good idea on most hardware; is there anybody trying to run XEmacs in 16MB anymore? Or you could try (setq gc-cons-percentage 40). (Aidan proposed a patch to make gc-cons-threshold 8MB by default, but Ben basically said "I'm getting rid of gc-cons-threshold in favor of gc-cons-percentage, which gives constant amortized GC-ing", so nothing happened. But I don't know if gc-cons-percentage is fully implemented in 21.5.20 or not.) Raymond> Would it matter if I changed back to 3.1? Would you be Raymond> interested in the results? I don't think it's worth it. The only optimization I've heard of that actually produces measurable speedup is -Os ;-). You might try that. -- School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Ask not how you can "do" free software business; ask what your business can "do for" free software.