Hi, Here is a patch I'm composing: http://www.hippo.ru/~hvv/tmp/abi-l10n-patch.gz First, all changes are xp or for Unix. This patch uses an instance of XAP_EncodingManager for all needed services with respect to translation of characters to/from native charset, getting windows' language code, arguments for various TeX packages, etc. Currently, the implementation of the class that derives from it on Unix has values and logic hardcoded form russian language and koi8-r charset. I will be tweaking that file in tomorrow - but everything beside that file is finished as is working fine. I repeat - all files I would be changing is only implementation of virtual functions of class derived from XAP_EncodingManager. Most notable changes: * Everything is working fine as user might expect :) * I tried to document key changes and their reasons. First read xap_EncodingManager.h to get the idea, and everything else after that. * AW loads only fonts with encoding equal to "native" encoding, ignoring other fonts with same family and style (i.e. with "wrong" encoding) * There are now 2 types of rtf user can export to - "plain RTF" (for export to Word97 and above and to StarOffice) and "RTF for old apps" (also understood by Word97 and above but as well as below and by wordpads and other stupid or ancient apps). Cut and paste is performed using "plain RTF". * I've tweaked export to latex - now much more unicode characters are converted to TeX equivalents. Also fixed export to TeX of non-latin1 chars. * My patch to automatically select color of toolbar icons (for icons that only use 2 or 3 colors) to match current gtk theme's text color. There are 2 RTF files exported (one for each format): http://www.hippo.ru/~hvv/tmp/f{old,cur}4.rtf - they correspong to RTFs saved with "plain RTF" format and "RTF for old apps". As you can see, \ansicpg and \fcharset now have correct params, and also \deflang is output. Also, for "plain RTF" plain equivalents of characters with unicode values > 0xff are output, i.e instead of \uc0\uNNNN AW will write \uc1\uNNNN\'NN (this makes files understandable by Wordpad from win95 and Words with version < 8). On import, parameters of \ansicpg and \deflang are respected. In general I don't recomment you to commit it :) Let me finish proper implementation of "backend" (to make them not russian-oriented). But please read it and make all comments/suggestions/corrections - I will probably be able to fix them. I will announce again when I've finished everything and consider everything complete. Best regards, -Vlad
