Re: Looking for new language features (re-elaboration)

William Rayer <lingolanguage@hotmail.com> wrote:
>
> What I am asking is, [... are] there even higher level features
> that we could put into a new language that would produce a similar
> order of improvement?
>
> Or are there "wish lists" of things that people would like to see in
> langauges? Again I realize you can't design a language just by joining
> up lots of nice features, and I realize the harder task is to make the
> features work together properly. But this post is about looking for
> ideas, not implementations.


Well, there is.


I'd like to see more provision for expressing proofs about program
properties, and in a way that the "average programmer" can at least
read and understand.


For example, it would be nice if the compiler could prove that the
code is consistent with its assertions. That way, the compiler would
be able to remove the assertion checking code automatically. Another
important property would be the "abstract constness" of a function
parameter (i.e. any changes to the parameter will be invisible to
callers; the function is then called "referentially transparent",
allowing the compiler to postpone or advance the function call as is
expedient, or to cache the function result, without changing the
program's semantics).


This would require a very expressive assertion language, e.g.
expressions in a functional language, and sprinkling the code
liberally with assertions to give the compiler a chance to deduce the
desired properties.


(I'm aware that there are theoretical and practical problems here, but
you were asking for ideas, not implementations <g>.)


I'm also dreaming of a language that's as useful for algorithmic
programming as for glueing components together. Newer scripting
languages look quite promising, but they aren't the "real thing" yet.


Regards,
Joachim
[I'd like a language that could compile the assertions so I don't have
to write the code. -John]

Return to the comp.compilers page.
Search the comp.compilers archives again.