compiling es

[John (_You_ hide, they seek.) Mackin <john@civil.su.oz.au>]

Well, one of the nice things about rc is that it wasn't written in
ANSI C (`So What?'), but rather in real C with prototypes, so that
if you ran it through that neat unproto device it would just compile.

I'd really like to vote in favour of es coming around to that point of
view.  `Use gcc' is not, in general, a satisfactory solution.  gcc was
designed and built with CISC machines in mind.  Sure, if you are compiling
for a 68000 or a VAX, it might well produce better code than any compiler
besides Ken's new one -- but who _is_, these days?  These are RISC days,
and while I am not prepared to flatly state that it is not possible for
gcc to generate good code for (say) the MIPS architecture, in theory
that is, the fact of the existing matter is that MIPS's compiler
produces _much_ better code than gcc (as well as being more reliable
and generally better-behaved).

Please?

That said, luckily Noel had compiled gcc on my machine already (_I_
wasn't gonna do it!) so I was able to compile es; further comments
pertain to gcc 2.2.2 running on Ultrix 4.2.

I needed to re-order #include's in two places in order to stop
warnings about NULL being redefined -- some include files are
smarter than others.  In es.h, <string.h> must come before
<stddef.h>; and in which.c, <sys/param.h> must come before "es.h".

OK,
John.