Hrvoje Niksic <hniksic(a)srce.hr&gt; writes: Could you be more explicit? I couldn't find the referenced message (searching on the subject and on the Message-ID), sorry. Hrvoje> I've now looked around, and found the answer to this in Hrvoje> `specifier-matching-instance'. I've even implemented the Hrvoje> display_table_instantiate() function and made the Hrvoje> appropriate arrangements for the fallthrough to work, but Hrvoje> I'm afraid the performance impact is tremendous -- no Hrvoje> caching is possible, and specifier_instance() stuff needs Hrvoje> to be called for every displayed character. Unacceptable. The documentation for `specifier-matching-instance' is ambiguous as to whether specifier_instance stuff() is called or inlined, but it does say "this is what redisplay does" (under Mule). Is that false? If it is true, why is calling of specifier_instance() stuff unacceptable, as compared to the current implementation of Mule (that is already known to be unacceptable)? Or are you saying you've isolated the cause of unacceptable Mule performance? That would be cool, especially if it can be fixed! -- University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Institute of Policy and Planning Sciences Tel/fax: +81 (298) 53-5091 __________________________________________________________________________ __________________________________________________________________________ What are those two straight lines for? "Free software rules."