johnl@ima.UUCP (11/08/85)
/* Written 4:59 pm Nov 7, 1985 by macrakis@harvard in ima:net.internat */ > Representing multiple character sets with shifts has been suggested on > this list. Roy Smith points out that this would cause great > difficulties in using Unix tools, for instance: > > Guess what breaks: > > Tail -- you [have to] track the shift commands. > > Grep -- should grep track the shift commands... > > Do you now need a shift-optimizing filter ...? Sounds to me like it's a bad idea to have shift sequences with global effect, which is not too surprising. If we declare that each line has to be self-describing, with appropriate shift sequences at the beginning if it is in other than the standard character set, that makes it entirely possible to make tail, grep, and such work. I think that we need to examine our conventions and assumptions about our conventions very carefully, and if we do, we can come up with a set that makes it possible to get work done without having to rewrite every bit of code from scratch. John Levine, ima!johnl
franka@mmintl.UUCP (Frank Adams) (11/15/85)
In article <125100002@ima.UUCP> johnl@ima.UUCP writes: >Sounds to me like it's a bad idea to have shift sequences with global >effect, which is not too surprising. If we declare that each line has to >be self-describing, with appropriate shift sequences at the beginning if >it is in other than the standard character set, that makes it entirely >possible to make tail, grep, and such work. OK, but I think you still want to be able to put a global override for the entire file at the beginning. A rule that says shift sequences not at the beginning of the file affect the current line only seems like a reasonable compromise. Frank Adams ihpn4!philabs!pwa-b!mmintl!franka Multimate International 52 Oakland Ave North E. Hartford, CT 06108