Re: Char-related crashes (hopefully) fixed

mb> Conceptually, it should be trivial to have a --with-mule build mb> that that acts in a Latin-1 way -- provide a function that mb> makes sure that only 'binary is used for all encoding/decoding mb> operations. This way there will be no unexpected surprises mb> decoding tar.gz files or whatever. This should probably be mb> implemented using a language environment. Maybe it should mb> even be the default ???!!?? Welcome, Martin! A lot of this was hashed over a couple of weeks ago (on a CC list), I put forward a pretty detailed proposal which Ben reviewed. As soon as I finish grading final exams I'll get back to that. You were on the CC list, so you should have a copy; if not I can send you a digest of the relevant discussion. Olivier Galibert has a pretty detailed plan for dealing with the `unibyte' issue. No naming problems, it will all be totally automatic. As for new architecture, Ben's pretty much signed off on that, as has Steve. We can call it Mule or not; at a high level, it will have to integrate with existing Mule APPs[1] for backward compatibility, but internally we are substantially diverging from FSF already. I think it's time for a clean break _if_ Ben and Olivier have the resources to follow up on the plans they've already suggested. Ben at least has a lot of other things in mind I know, but he seemed to put pretty high priority on Mule when we were discussing it earlier. Footnotes: [1] Application Programming Practices. I will not dignify the current mess with the name "interface". -- 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."