bug in the class browser

[Alexander Uwah]

On Wed, 16 Jun 1999 you wrote:
> > is already a special case in the parser code for one such name. Although my
> > grammar/parser knowledge is a bit rusty I am sure that parsing:
> > 
> > class +<identifier>  ( : <base_class_specs> {  | {  )
> > 
> > doesn't break LR(1) and that yacc can be made to only retain the last
> > identifier as the class name (apologies for the BNF, thats a bit rusty too).
> > 
> > This would at least make two people happy, the original poster and me :)
> 
> The thing is that I don't use yacc. This is not needed and will only complicate
> thing since I need a more or less complete c++-grammar for that.
> 
> I'm just using a tokenizer and a stack of lexem right now. 
> 
> /J

Hello Jonas,
Is it not possible to get a complete c++-grammar somewhere ?  I mean, it looks
like the parser is a very important part of kdevelop and I think it is worth
the effort if this part is made as robust as possible. It´s happened often to
me that I tried to include some already written code and kdevelop crashed (I
think while parsing).
Error (Typos) handling can be made strict, so that if there is a typo, a
corresponding warning can be issued without letting the parser "make a guess".
Or is that a bad idea?

Alexander