From: <don...@is...> - 2002-01-07 01:48:30
|
Sam Steingold writes: > > For dealing with standard internet protocols I long ago gave up trying > > to use characters in clisp and moved to bytes. At least for reading. > > this is unfortunate. > what is the reason? > maybe we could patch CLISP to your tastes? :-) I don't know whether it's unfortunate. As I recall, there were two reasons. One was the end of line issue and the other was errors with character sets that had no character defined for the bytes that came in. Notice that when you write a server you don't have a lot of control over the bytes that others send you. I recall being told, (just as you said recently on this list), that if you want to deal with CR and LF as different things then you're working at too low a level for characters. I suppose it's possible that enough stuff has been added over the last few years that I could now do all I want with characters, but I recall that I did waste (in retrospect) a lot of time trying and failing. My original code using characters did work in Allegro CL, but I think that's because the default "character" set there was actually no different from bytes, i.e., there were 256 of them, each represented by a different 8 bit byte. Perhaps clisp has such a character set, but it was at least not the default when I was trying to write this code. In fact I recall trying different character sets. Perhaps others who have written such software can confirm or refute my impression that both the problem and solution were widespread. |