GSOC 2012

[Sven Brauch]

> Fortunately, there is no need to break everything. KDevelop is modular
> enough to handle two parsers fortunately. Also this does not have to
> live inside KDevelop's repository until it is ready.

Still: This is a project which needs to be planned in a timeframe of
years, regarding the current manpower. And until it's really really
finished, it provides no real benefit. It's luxury, and we don't have
the resources for luxury. I don't know, that's just what I think.
Oh, and the other thing is also a valid point: clang isn't easily
available on a ton of platforms right now (I think most recent linux
distros don't really ship it, nor do they even have it in their
repositories). Thus it introduces yet another hard-to-install
dependency.

> As for the 99% thing, I use KDevelop mainly for kernel hacking and
> it's far from being 99%. I try to fix things here and there, but there
> will always be corner cases for which a full-compliant parser will be
> needed.

That's to be expected with a syntax as over-complicated as the C one,
I guess... still, if the DUChain code doesn't handle a particular
case, the parser alone won't be of much use.

Greetings