Path: utzoo!mnetor!uunet!husc6!bbn!rochester!crowl
From: crowl@cs.rochester.edu (Lawrence Crowl)
Newsgroups: comp.lang.misc
Subject: Re: Perfect language features: how many languages?
Message-ID: <6562@sol.ARPA>
Date: 5 Feb 88 16:00:32 GMT
References: <3928@ames.arpa> <2400001@otter.HP.COM> <960@ssc-bee.ssc-vax.UUCP> <1535@uoregon.UUCP> <1496@mips.mips.COM>
Reply-To: crowl@cs.rochester.edu (Lawrence Crowl)
Organization: U of Rochester, CS Dept, Rochester, NY
Lines: 35

In article <1496@mips.mips.COM> hansen@mips.COM (Craig Hansen) writes:
>There is a basic problem in language design: a single language that is good
>for all purposes will be very complex.

Only if you insist that the libraries be considered part of the language or
you insist on languages which do not make the notions of type and procedure
available to the programmer.

>A language designer, working alone will be unable to design such a language,
>and it isn't at all clear that a committee of language designers would bring
>any more effective intellect on the problem.

I believe that only a single language designer can achieve such a language.
Such a language requires deep coherency between all aspects of the language,
something that is unlikely in a committee.  This does not mean that the
designer works alone---others provide feedback, but the designer is responsible
for all decisions.

>Now a little language, by itself, only works in a limited problem domain, and
>trying to pound nails with a screwdriver is a waste of effort. UNIX works well
>because it encourages many small tools and languages to be used together,
>using pipes and shell scripts as the glue between the little pieces.

People must also learn an incredible variety of little languages.  Most of the
little Unix tools work OK with each other, but they do not work well.  How many
ways are there to delete something in Unix?  How many mechanisms for changing
the interpretation of shell paramters are there?

Unix requires far too much special case, "wizard" knowledge.  I am not arguing
that anything is better than Unix, only that its many language approach
introduces far too many inconsitencies to be the desired solution.
-- 
  Lawrence Crowl		716-275-9499	University of Rochester
		      crowl@cs.rochester.edu	Computer Science Department
...!{allegra,decvax,rutgers}!rochester!crowl	Rochester, New York,  14627