Re: Mule autodetection is crap

SJT> Well, as I suppose you know, what Ben has is mind is a statistical SJT> detector that (eg) can distinguish EUC-JP from EUC-TW or EUC-KR SJT> (although the really important case is the ISO-8859-X mess and the SJT> various non-conforming sets like KOI8 and the Windows 12xx sets). I can say that there is an extremely good scheme for statistical detection of various Russian (really Russian, not Cyrillic) encodings, done by S. V. Znamensky. I tried it, and it works really wonderful, allowing even "twice-encoded" text which is seen occasionally. I thought of adding something like this to XEmacs. Now if there is a common infrastructure for this, I'd be glad to help in that area. SJT> But the design is completely new, so we need to retune it. Also SJT> there seem to be some bugs in coding priorities. Uhm, I'm now playing with current XEmacs-beta. It recognizes my ~/.xemacs/init.el as UTF-16, and does not let me to change the encoding with "C-x RET f koi8-r RET" (but "C-x RET c koi8-r C-x C-f" works). The file itself is mostly ASCII, with two strings in Russian inside (near the end of the file). Are you interested in such bug reports, and if yes, should I send the file or what? Other files it at least detects as "Raw". set-language-enviroment Cyrillic-KOI8 does not help at all. --alexm