Variable-Length Array Parameters in NCEG C

[Conor O'Neill]

Richard Stallman writes:
>Compiler writers should do "the work" once--but they should do the
>right work, not the wrong work.  They should implement a language that
>has a clean definition.  A language whose features work all the time,
>not just most of the time.  A language whose meaning is very easy to
>predict.

He then adds, in another article:
>>            b[n*n, 3*m + p]
>Using the syntax I proposed, this does not cause a problem.
>The prototype would look like this:
>
>void a (int n; int m; int p; float b[n*n, 3*m + p], int n, int m, int p);

I would say that this most definitely _does_ cause a problem.
This prototype is extremely unclear about what order the parameters are.
I would not consider this to be a "language whose meaning is very easy to
predict", and certainly not a language which is easy to read.

---
Conor O'Neill, Software Group, INMOS Ltd., UK.
UK: conorainmos.co.uk		US: conorainmos.com
"It's state-of-the-art" "But it doesn't work!" "That is the state-of-the-art".