Path: utzoo!mnetor!uunet!lll-winken!lll-lcc!lll-tis!ames!amdcad!light!bvs
From: bvs@light.uucp (Bakul Shah)
Newsgroups: comp.lang.c
Subject: Re: Conformant Arrays in C
Message-ID: <1988Feb24.165307.4938@light.uucp>
Date: 25 Feb 88 00:53:04 GMT
References: <42529@sun.uucp> <7297@brl-smoke.ARPA> <676@cresswell.quintus.UUCP> <2336@umd5.umd.edu> <684@cresswell.quintus.UUCP>
Reply-To: bvs@light.UUCP (Bakul Shah)
Organization: Light Systems, Mountain View, CA
Lines: 25
Keywords: ANSI C, C++
Summary: Leave it upto C++

Multi-dimensional arrays are a useful feature and David Hough / Richard
O'Keefe's proposal of conformant arrays is usable & implementable.  But
the proposed conformant arrays are not elegant and, in my opinion, it is
a *bad* idea to add them to C (atleast as bad as adding structure
comparison or nested procedures or name-your-favorite-non-C-feature).

Such things are best left for C++.  No new features are needed to
implement multi-dimensional arrays in C++, Chris Torek's ``row vectors''
would do just fine AND, and the machinery will be hidden (what O'Keefe
wants in a >sensible< language).

If they are added to C, C++ will be forced to carry them (thereby
introducing an unneeded hack)  OR, C++ will *not* be upward compatible
to C.  Since conformant arrays do not exist in C today, only new code
will use this feature (if added).  Why not use C++ or Chris Torek's idea
for new code?

C just doesn't have the bandwidth to do many sensible things.  Attempts
to increase this bandwidth in an elegant way is extremely difficult, if
not impossible.

My two cent's worth.
-- 
Bakul Shah

..!{ucbvax,sun}!amdcad!light!bvs