Re: linux headers and C++

Dancer (dancer@zeor.simegen.com)
Wed, 07 Jul 1999 00:11:53 +1000

Jes Sorensen wrote:
>
> >>>>> "Jamie" == Jamie Lokier <lkd@tantalophile.demon.co.uk> writes:
>
> Jamie> Albert D. Cahalan wrote:
> >> Linux already uses more than a C compiler offers. The best answer
> >> would be to merge gcc and g++, using a config file to enable the
> >> features that Linus approves of. Linux could use "C+" then.
>
> Jamie> One of the reasons Linus stopped using g++ on the kernel was
> Jamie> the quality of the generated code. I gather g++ generates much
> Jamie> better code now.
>
> Jamie> It'd still be important to disable exceptions (for example).
> Jamie> And templates remain dangerous bloat-generating territory.
>
> And templates, and operator overloading etc etc.
>
> C++ is much harder to maintain because there are a zillion obscure
> features people can use in there. It means that the people who
> maintains the kernel needs to understand these things as well thus we
> suddenly reduce the number of potential kernel programmers to a tiny
> fraction of what we had before.

I do a lot of stuff that involves attaching C++ code to C code. Really,
it's the C++ coder's responsibility to insulate other subsystems, if
they're bolting C++ onto a C framework. Likewise, when a C programmer is
bolting C code onto C++, it's his responsibility to ensure that the C++
base is unaffected.

I've seen C++ code that plugs into the kernel (it was a driver module,
in fact). No changes to the kernel required, just a little care from the
C++ coder.

So...why are you guys fighting? Is it just on principle, or is there
some valid objection that hasn't been raised yet?

D

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/