On Fri, Oct 24, 2008 at 08:11:50PM -0700, John Bergamini wrote:
> 1. I was unaware that C-- needed reviving. I thought it was very much
> under active development.
>
> 2. I am not an expert, but C-- appears to be the perfect thing for my
> purposes. My purpose is to implement the language Grok32`.
It looks good for my purposes, too, to the extend that I was thinking of
replacing my existing code generator that generates JANUS code with one
that generates C-- code. (An alternative is to translate JANUS to C--)
At
the moment, JANUS seems even deader than C--. It has no machine-code
generators for anything more modern than an IBM 370.
I suspect that C-- could benefit from a comparison with JANUS. The
semantics of JANUS are well-thought out and well-specified -- to the
extent that a decent portable mathematical functions library was
possible. JANUS also addresses procedures-within-procedures and the
ensuing variable-access issues. It lacks crucial features like
efficient tail-calls, though.
>
> 3. I find it particularly troubling that nobody is finishing work on
> the PowerPC or derivative processors. Why?
Probably the same reason the project seemed dead -- lack of manpower.
>
> 4. What is the evidence that C-- is dead?
(a) the paucity of messages on the C-- mailing list since 2006 Feb
17, and that one was spam.
(b)the status report from Norman Ramsey on 2006 Dec 15,
https://cminusminus.org/lists/pipermail/cminusminus/2006-December/000008.html
after which nothing much appears to have happened.
Since I posted the message you are replying to, though, an off-list
reply I received indicates that the project will likely come to life
soon.
I'd bet on it by starting to write actual code that uses it, except
that I'm crazy-busy until December.
-- hendrik
>
> John B
>
> --- On Thu, 10/9/08, Hendrik Boom <[EMAIL PROTECTED]> wrote:
> From: Hendrik Boom <[EMAIL PROTECTED]>
> Subject: [C--] Is it worth reviving C--?
> To: [email protected]
> Date: Thursday, October 9, 2008, 11:15 AM
>
> I have a not-quite-complete Algol 68 conmpiler, and am wondering about a
> new code generator. The old one doesn't seem all that useful these days
> because it generates IBM 360 code. I'm currently looking at C-- and llvm
> as back ends. C--, as distributed, seems more hoepitable to garbage
> collection and such, but llvm seems to be more actively maintained, and
> seems a lot larger.
>
> C-- also does not seem to have an AMD64 code generator, nor one for the
> ARM (an increasingly important atchitecture)..
>
> I'm wondering if there is enough potential interest in C-- to be worth
> reviving it.
>
> I couldn't develop it or maintina it single-handed, but if others are
> interested, I could certainly maintain a development repository. I'd
> probably pick a distributed repository, like monotone, since that seems to
> be a good combination of paranoia about system corruption and freedom to
> tinker at will.
>
> Are there others interested? This would definitely be done on a
> free-software development model, consisting of volunteers all over the
> world.
>
> Opinions, please. Potential users, please speak up. Potential
> contributors, likewise.
>
> I have a decision to make. C-- only if it is supportable. Or llvm, which
> doesn't seem to be as good a match to my front end, but has lots of
> implemented targets architectures.
>
> Has anyone investigated c-- to llvm, by the way?
>
> -- hendrik
>
> _______________________________________________
> Cminusminus mailing list
> [email protected]
> https://cminusminus.org/mailman/listinfo/cminusminus
>
>
>
>
_______________________________________________
Cminusminus mailing list
[email protected]
https://cminusminus.org/mailman/listinfo/cminusminus
