Nicolas, I see your point about specifying actual dot patterns. Perhaps the translation table could be set up to allow both dot patterns and ASCII. I'll try to find the French Braille code. It looks like a good test case. It turns out that the format of the translation tables in nfbtrans is similar to the one we have been discussing: number (pattern matching method), source string, replacement string. However, nfbtrans appears to be the product of many hands, with bits and pieces added from time. It is also very inefficiently coded, with redundant if statements. Considering my experience in program maintenance and in writing translators for other purposes, I think it would probably be less time-consuming to just start from scratch. Though I wouldn't actually be starting from scratch, because I already have a translating routine that might be modifiable. John Computers to Help People, Inc. http://www.chpi.org 825 East Johnson; Madison, WI 53703 ----- Original Message ----- From: "Nicolas Pitre" <nico@cam.org> To: <blinux-list@redhat.com> Sent: Monday, November 26, 2001 18:09 Subject: Re: brltty and simultaneous grade 2 translation > On Mon, 26 Nov 2001, John J. Boyer wrote: > > > I like Dave's suggestion of putting the number first in the rules. This > > number actually indicates a pattern-matching method. So the algorithm would > > be more than a simple pattern replacement. I know French. Where could I find > > their Braille code, preferably on the Web? > > Don't know. I learned compressed French braille a couple years ago. > Although I still can read it more or less with surrounding context, I fear I > wouldn't be able to write it anymore. There are the pretty generic rules > which are sensible, but you must also account for the hundreds of ad hoc > rules. > > > I think that a direct dot pattern representation would not be as clear as > > using the ASCII characters that are used to represent Braille dot patterns > > in the particular language. Don't the text-translation tables already take > > care of translating to whatever characters a particular display needs to > > represent a particular dot pattern. > > Well... not necessarily. At least not in French, especially when it comes > to computer braille. French braille is quite well defined for > paper literature, but computer braille is a much random matter, especially > for non alphabetical characters. A simple dot "." may translate to > different dot pattern depending on your taste. I even got my own braille > table to use with BRLTTY as I couldn't stand the other available tables > which seemed not natural to me. > > Therefore it would be much simpler, at least for the French braille > compression, to just tell BRLTTY: replace each "ation" sufix in a word by > dots 3-4, "ent" sufixes by dot 1-2-6, and so on, since that's actually how > it is defined for paper document and not fiddle with the current braille > table for the non translated case. Still I may wish to read compressed > English text without having to change my own braille table for something I > couldn't understand just because the braille compressor was defined with a > particular braille table in mind. > > > Nicolas > > > > _______________________________________________ > > Blinux-list@redhat.com > https://listman.redhat.com/mailman/listinfo/blinux-list >