KDevelop 1.0beta3 patch - Objective-C support added

[Andrew Morton]

Jonas Nordin wrote:
> 
> > Yes, there are other ways in which YY_USES_REJECT gets turned on.
> > However the beta2 version of tokenizer.l doesn't use them - I tried
> > removing '%option yylineno' and the flex output no longer defines
> I was too lazy to do my own line number counting. Since there aren't too many
> rules in the tokenizer.l it should be doable. I personally have 0 time to
> allocate to this right now. Please feel free to fix it. :-)

My time isn't looking too good for the next few weeks either, I'm
afraid.

Richard, are you there?  Have you time to look at the mega-comment
problem?

> > While I was in there I significantly simplified some of the rules in
> > tokenizer.l. Probably not very significant though.
> The complicated ones where "borrowed" from someones c++-grammar. :-)

The wonders of opensource.

Jonas,  I'm interested in the persistant storage stuff in the
classparser.  Is this designed to improve startup time for large
projects? (Mine's 70 megs).

Do you plan to go ahead with the implementation of this?

The reason I ask: I have implemented an object serialisation framework
which should do the job.  It's very well tested, handles base classes,
morphologically correct, etc.

I've been meaning for years to extract it from the rest of my reuse
library and my compiler, redo the template instantiation and to document
the thing, but nobody's given me the motivation...

Have a look at http://www.uow.edu.au/~andrewm/cs++.html and
http://www.uow.edu.au/~andrewm/cs++.tar.gz

Implementation is in

	AMROOT/CORE/inc/*Packer*
	AMROOT/CORE/src/*Packer*

Example code in AMROOT/CORE/examples/TestClasses.h and packers.cc

It needs documenting...