Re: [C--] C-- development

On 03/14/2013 10:00 AM, Simon Peyton-Jones wrote:
> Yes, C-- as a standalone project is moribund in the sense that no one is working on it. But I think there were some interesting ideas in C--, notably concerning the runtime API for exception handling and garbage collection.  I'm not sure of the extent to which these ideas show up in LLVM, but that's just my ignorance.
>
> However, in effect, LLVM is now the incumbent on the "portable assembly language" patch -- and more power to them!  There is a critical mass of people and companies gathered around LLVM doing great things.  One may argue (and some of you kindly have) that C-- does something better or more elegantly, but it's hard to argue with LLVM's momentum.  And that momentum is a good thing, because it motivates people to do the sweaty work of porting to new platforms etc. In effect, a "portable assembly language" HAS to have critical mass to meet its design purpose.
>
> That said, I think C-- is wide open to people jumping in and "owning" it themselves, a research/development project.  Just go for it; Norman and I would be delighted.
>
> GHC (the Glasgow Haskell Compiler) indeed uses C-- as an intermediate language, via its intermediate data type called 'Cmm'.  There's an external form too in .cmm files.  But in both cases it is somewhat adapted to the GHC setting, and so is not intended to be a standalone language useful as a backend for many things. (It may be, but we've felt free not to make that a constraint.)  As Norman says, it's an "extended subset"!
>
> We don't use qc--; rather there's a Cmm -> native code compiler in GHC.  And, in fact a Cmm -> LLVM generator that then uses LLVM as a back end.
Will GHC benefit from separating Cmm -> LLVM/native code generator out 
as a standalone backend?

The Cmm backend can be built and tested independently from GHC 
front-end, and it can also be used by other functional languages.

Ning
_______________________________________________
Cminusminus mailing list
[email protected]
https://lists.eecs.harvard.edu/mailman/listinfo/cminusminus