Re: GCC is re-implementing in C++ and C discarded
 

William Ahern <william@wilbur.25thandClement.com> wrote:

> Anonymous <nobody@remailer.paranoici.org> wrote:
> > "Steven G. Kargl" <sgk@REMOVEtroutmask.apl.washington.edu> wrote:
> >
> > > The GNU Fortran frontend is written in C. Some of the GNU
> > > Fortran runtime is written in m4. The m4 files when processed
> > > generates Fortran code. The Fortran code is then compiled by
> > > GNU Fortran.
> >
> > OMG no. You didn't just write "Some of the GNU Fortran runtime is written in
> > m4."
> >
> > Oh yes you did. Just another reason not to use gnu's shitty toolchain. How
> > do you mother****ers sleep at night?
> >
> > Goodbye to gcc's grotesque bloat and "i was drinking that night" design,
> > goodbye to Stallmanesque Marxism, hello clang/LLVM!
>
> Let me get this straight: you call GCC marxist because, unlike clang/LLVM,
> it doesn't have the quality of language purity?

No, gcc is Marxist because Stallman is Marxist. Sorry for any confusion.

> Because that's basically what people seem to like about clang/LLVM: that's
> it's "pure" C++, which somehow magically makes it better designed. (It may
> or may not be better designed than GCC, but C++ doesn't bestow that
> quality.)

I agree with you.

>
> You also seem to forget that M4 and Fortran share a common history.

No they don't. Fortran was developed at IBM on IBM machines a very long time
ago, around 1957. This was well before there was such a thing as UNIX or m4.

> M4 was originally used to provide control constructs which early Fortran
> lacked.

I don't know if you're referring to RATFOR, I don't know how that was
implemented. But that was in UNIX and IBM didn't need it. There were people
using FORTRAN happily then and now on those machines. The compilers were
totally different from anything that came later on other systems. m4 is
really not pretty. IBM had and has today much better (as in more powerful,
more expressive, and more simple to use) preprocessors than are available on
UNIX. Those can be used for FORTRAN as well, with a bit of effort.

> I seem to recall "meta-programming" being all the rage, although
> apparently it's less useful if the name of the meta-programming language
> isn't synonymous with the target language. Because, you know, then it's
> not pure.

Metaprogramming may have been popular on minis, I don't know. But as far as
I know it was never popular on large systems because large systems work was
always about getting stuff done in a simple and high performance way rather
than inventing cutesy names for broken ideas and then turning them into fads.
And anyway I wasn't arguing purity when expressing my loathing of gcc and
everything related to it. I just hate bloat, convoluted code and communism.
Most other stuff doesn't bother me that much ;-)

>