home *** CD-ROM | disk | FTP | other *** search
- Xref: sparky comp.arch:10471 comp.lang.misc:3529
- Newsgroups: comp.arch,comp.lang.misc
- Path: sparky!uunet!utcsri!skule.ecf!torn!watserv2.uwaterloo.ca!watserv1!mhcoffin
- From: mhcoffin@tolstoy.uwaterloo.ca (Michael Coffin)
- Subject: Re: Hardware Support for Numeric Algorithms
- In-Reply-To: preston@miranda.cs.rice.edu's message of Thu, 5 Nov 1992 17:09:34 GMT
- Message-ID: <MHCOFFIN.92Nov6142412@tolstoy.uwaterloo.ca>
- Sender: news@watserv1.uwaterloo.ca
- Organization: Dept. of Computer Science, University of Waterloo
- References: <1992Oct22.163955.12161@newshost.lanl.gov>
- <GLEW.92Nov4072629@pdx007.intel.com>
- <MHCOFFIN.92Nov5075939@tolstoy.uwaterloo.ca> <Bx96Bz.E2@rice.edu>
- Date: Fri, 6 Nov 1992 19:24:12 GMT
- Lines: 46
-
- In article <Bx96Bz.E2@rice.edu> preston@miranda.cs.rice.edu (Preston Briggs) writes:
- >
- > mhcoffin@tolstoy.uwaterloo.ca (Michael Coffin) writes:
- >
- > >I don't disagree that occasionally gotos are cleaner than the
- > >available alternatives, but I wouldn't call Knuth an arbiter of good
- > >taste in programming style. On the one hand, TeX the program is
- > >clearly written and contains some beautiful algorithms. On the
- > >other hand, TeX the language is a wretched mess---it's almost
- > >impossible to write clear code in TeX. So I'd say Knuth's record on
- > >programming style is very mixed.
- >
- > I disagree. It's easier to read and write programs in TeX than
- > in any other typesetting language around!
-
- It's a matter of taste I suppose, but I disagree. Except that the
- names are a little nicer, I prefer even troff macros, to say nothing
- of, say, Postscript.
-
- > Isn't your complaint similar to saying
- >
- > Makita builds terrible tools because this power planer
- > makes a terrible lathe.
-
- I don't think so. I just think that simple things ought to be simple
- to express. I don't find that to be true of TeX. Often simple things
- require bizarre contortions to accomplish.
-
- I think Knuth thought hard about specifying the output of a typesetter
- and did a good job on that aspect of the language. Macros work pretty
- well for output; it's sort of natural to think of macros `expanding'
- to the desired output by repeated rewriting. (Even here there are
- complications, especially the two-phase expansion: the `mouth' and the
- `stomach', as he puts it.) But he didn't design good features to
- parse the input, which is the flip side of a language for typesetting.
- The code required to parse even moderately complicated input---a
- context-free grammar, for example---is absurd. Contrast that with
- troff's eqn, where the input format is specified very naturally using
- appropriate tools like lex and yacc.
-
- -mike
-
-
-
-
-
-