npn@cbnewsl.att.com (nils-peter.nelson) (01/08/91)
I've received a proposal from Keizer, Simonsen and Akkerhuis for 2 character representations for the additional 128 characters in ISO 8859-1. Dennis Ritchie had already implemented a competing convention for Research UNIX for keyboard entry. For example, for the open French quote, or left angle quote mark, position 10/11, K/S/A suggest \(Fo, Ritchie suggests \(<<. If you have such a key on your keyboard, this is not an issue; if you don't, is there a common digraph in existence? Viz., does everyone type "dead-key char1 char2" the same way, or is it hopelessly local. To the point, are there any more proposals out there for how to represent 8 bit characters in a 7 bit subset? mail to npn@mhuxo.att.com. My goal is to have the troff convention be the same as the keyboard convention most familiar to people, if such a thing exists.
yfcw14@castle.ed.ac.uk (K P Donnelly) (01/09/91)
npn@cbnewsl.att.com (nils-peter.nelson) writes: >To the point, are there any more proposals out there for how to >represent 8 bit characters in a 7 bit subset? >My goal is to have the troff convention be the same as the >keyboard convention most familiar to people, if such a thing exists. Surely by far the most common keyboard convention for ISO 8859-1 is that on used VT320 terminals, in which for example you generate the 8 bit character "half" type pressing "Compose-character" then "1" then "2". Some other examples are: quarter 1 4 a-acute a ' or ' a A-acute A ' or ' A pound sign L - or - L French quote marks < < > > suberscript 2 2 ^ degree sign 0 ^ plus or minus + - Any user's manual for a VT320 or compatible gives full details. Kevin Donnelly
keld@login.dkuug.dk (Keld J|rn Simonsen) (01/12/91)
yfcw14@castle.ed.ac.uk (K P Donnelly) writes: >npn@cbnewsl.att.com (nils-peter.nelson) writes: >>To the point, are there any more proposals out there for how to >>represent 8 bit characters in a 7 bit subset? >>My goal is to have the troff convention be the same as the >>keyboard convention most familiar to people, if such a thing exists. >Surely by far the most common keyboard convention for ISO 8859-1 is that >on used VT320 terminals, in which for example you generate the 8 bit >character "half" type pressing "Compose-character" then "1" then "2". Well, well. I did some other work on this, and also had a look at the VT320 names. What I wanted to do was to have ASCII encodings of all the ISO 8859 character sets and also other character sets. I found that the VT320 codes were fine for ISO 8859-1, but when all the parts of ISO 8859 (there is about 10 parts) should be coded, there were conflicts in the naming. I now have a set of more than 1300 character names in 2 character ASCII (or actually invariant ISO 646), which is used for definition of POSIX locales and used in email. I also have tables of the encoding of about 60 character sets with these two-char names. Unfortunately these names are incompatible with Ossanna/Kernighan titroff. And thus the names in K/S/A are incompatible with this extended list. And I was involved in both lists.... The 1300 character list was a lot bigger than the K/S/A list and therefore had to be designed more consistently and carefully. One fundamental design decision was to shift around the letter position in letter names, so titroff *a is now a* in the 1300 char list. The 1300 char list includes the following: extended latin, greek, cyrillic, hebrew, arabic, mathematics, hiragana, katakana, bopomofo. Two ideas for this: 1. A titroff list of the 1300 characters (or the like) could be done - which could be as compatible as possible with the big list I have now. 2. the new titroff could have a specification of input character set - thus more markets could be opened up for this product. The 1300 char list (and some code to handle it) is available by anon ftp in dkuug.dk:pub/ch.shar* Keld Simonsen