[Info-vax] C99 stuff (Re: The Road to V9.0)

[Stephen Hoffman]

On 2019-06-07 19:17:15 +0000, Phillip Helbig (undress to reply said:

> Back in the day, DEC's compilers were widely recognized as the  best, 
> and in many cases were the de-facto industry standard.  I'd like to see 
> that revived.  In particular, it would be nice if Fortran were brought 
> up to date.  Reasonable versions of Pascal, C, C++, Cobol, and Basic 
> would be nice.

llvm and clang and clang++ are a ~decade ahead of the OpenVMS compilers.

clang is far more capable than the DEC C compiler and far more modular, 
and yet better when paired with a competent IDE.  Continuous 
compilation of code right in the editor is... really quite surprisingly 
useful

Work on f18 / flang is underway in the community, and the results of 
that effort will undoubtedly and eventually be ported over to OpenVMS.

https://github.com/flang-compiler/f18

That f18 / flang porting work probably can't even start until after C++ 
is available natively, based on another recent discussion here.

Various of the rest of the DEC compilers will almost certainly continue 
to use John's GEM shoe, and for the foreseeable future.

But an effort to try to "outrun" llvm?  That will not happen at VSI. 
Not for the foreseeable future.  Why?  The whole of the VSI development 
team is probably smaller than the team that's actively working on llvm 
and related tools.  There's just not enough of a differentiation here 
for what VSI can invest in that compiler and tooling effort, and 
there's no shortage of other projects pending.  And no shortage of ways 
that OpenVMS could start adopting more of what llvm and clang and 
clang++ provide, too.

It'd be nice to see the DEC front-ends for the various languages 
open-sourced for use with llvm, but that probably won't happen until 
2021, and maybe never.



-- 
Pure Personal Opinion | HoffmanLabs LLC