This is the mail archive of the
mailing list for the GCC project.
RE: Converting GCC to compilation with C++
- From: "Dave Korn" <dk at artimi dot com>
- To: <gcc at gcc dot gnu dot org>
- Date: Tue, 13 Jul 2004 12:11:57 +0100
- Subject: RE: Converting GCC to compilation with C++
> -----Original Message-----
> From: gcc-owner On Behalf Of rd flowers
> Sent: 13 July 2004 11:09
> Please forgive me if what I say here is either obvious or horridly
> braindead. I am meaning well.
> I would HATE if the Gnu C compiler became insufficient to compile
> itself. A lot.
> But, we have extensions. We could have more (off by default) -- really
> intended for compiling GCC. Just add, incrementally and
> an extension or 3 that help us do those things that motivated
> this whole
> Or not.
> R. D. Flowers
There's one thing even more important than the issue of whether we have to
use C or a C++ compilers to build gcc, and that is that it should always be
possible to build gcc using the native/proprietary (non-gcc) tools that come
with a system, so that you can then replace those tools with gnu tools. If
the gcc sources start making use of gcc-only extensions, it soon won't be
possible to build it with any compiler other than gcc. That would be a
disaster; it would seriously impede the goal of getting open source
software onto proprietary platforms if there's no way of building the
compiler on them.
My take on the C-vs-C++ debate would be i) that C++ compilers are now
widespread enough that it's almost certainly no longer a barrier to entry if
we started using C++ code in the compiler, ii) that I'd have very serious
reservations for the long-term maintainability (and for the ability of
newcomers to get to grips with the source, already difficult enough) of the
entire project if we started to go bananas with tangled and multiple
inheritance hierarchies, templates, inanely overloaded operators, and all
the more obfuscatory features of the C++ language, iii) but that I'd not be
concerned if we were using a sensible subset of the language, and that iv)
it's not all that important one way or the another really, subject only to
the maintainability/portability/comprehensibility/entry barrier concerns I
mentioned earlier and the overriding need for gcc to provide an escape route
for those trapped by proprietary software.
Can't think of a witty .sigline today....