[Gpsbabel-code] re: c99 pervasiveness

[Robert L. <rob...@us...>]

A bit off topic, but we'll try this for a while...


> Are any of the compilers that are typically used to build gpsbabel C99
> compliant?

That's one of the questions I was trolling.  When I was watching the
compiler biz like a hawk (background: I used to be an active contributor
and maintainer of tools like GCC) I had the impression that most of
the guys had most of the support that was really useful to the world
at large.  Much of the work in the library and language was codifying
existing practice and borrowing small pieces from C++.  And since most
C compilers are really C++ compilers these days, that wasn't a huge
stretch.

The really wacky stuff (_Complex, _Imaginary, VLAs, etc.) doesn't seem
to be gaining ground, but it's not like it's a deal-breaker for us,
either. Though I have hunted geocaches where I'm convinced the coords
were stored in type _Imaginary and _Complex. :-)

> If you read comp.lang.c, there are many smart people who believe C99
> is stillborn.  Comments are made about the lack of C99 compilers in

Interesting.  I haven't read that group in a long time.

> lack of non-compliant compilers in existance five years after the
> release of C89/90.

Times were much more troubled then. C89 solved a lot of problems for
the industry, the bar was set lower, and I think there was more pent up
demand for a contract between coders and translators.

> I use many modern C90 compilers, and all of them optionally support
> "//" without supporting any other C99 or C++ features.  Not that
> gpsbabel could be compiled by any of those compilers...

[ scratches head ]
I've tried pretty hard to keep the project C89 with a little leaking in
of popular C99-isms like snprintf and // (both of which, as you observe,
have been industry practice for many years)   Are you saying we have 
portability problems beyond those?

RJL