As a conclusion, I think that one needs to build dictionaries which map trees into translated trees, where the source trees are as small as possible (paragraphs, or even sentences).
Yes, this is definitely the way to go. I thought some time ago that there had to be a way to write multilingual documents and keep them in sync. At the time, it seemed to me that one would enclose different language correspondents of one "unit" (e.g. one paragraph) in an environment (e.g. \spanish) and then selectively make visible the source and target language (english, spanish) for translation and only the target language for reading. In principle the typesetter is not slowed down by all the content which is only optionally displayed and not currently visible, right?
At any rate, I temporarily gave up the improvement of the translation process, because I fear it to be quite tricky and time consuming. If I had one month of time, then I could do a very nice job. But unfortunately, I don't.
It looks like all the infrastructure for the above mentioned "simple" translation approach, which overlaps with what is needed for revisions and annotations to a certain extent would be doable in scheme, just by defining the right environments. Regarding the wiki source format issue, I think that it would be valuable to have a compatibility mode which would work by writing html (therefore being compatible with current wiki engines) and ship the mathematics either as a png or as MathML. Trying to have wikis remain as much as possible browser-writable is a worthwile goal... Regards, Álvaro. _______________________________________________ Texmacs-dev mailing list [email protected] http://lists.gnu.org/mailman/listinfo/texmacs-dev
