Re: Parser base class (lalr1)

[Hans Aberg]

On 28 Sep 2005, at 21:02, Jan Nabbefeld wrote:

> Hello,
> On 28.09.2005, Hans Aberg wrote:
>
> > On 28 Sep 2005, at 14:37, Jan Nabbefeld wrote:
> > I use a C++ polymorphic hierarchy, with a reference count at the
> > base, in order to avoid unnecessary copying.
> > This approach has also the advantage of making use of C++ cleanup
> > The code above then becomes:
> >
> > cd_list: /* empty */ {$$=0;}
> >     | cd_list cd { $$ = new NRec(token, _CD_LIST, $1, $2, _CD); }
> > ;
> > [snip]
>
> That's a nice solution but unfortunately no option for me. As I
> mentioned, there is no way for me to change the semantic actions. My
> mission is to rebuild an existing C implemented scanner/parser in C++.
> The semantic actions have to be untouched (as far as possible).

It seems me that you want to put the functions in a separate  
namespace, but having to write out the namespace name in the actions.  
Have you tried using the C++ "namespace" and "using" features.

Also, if you put the tree* in the parser class, then the parser isn't  
pure. Do you want that? If you put it in the parser function  
argument, instead, then you get a pure function (or at least, when  
the C++ parser has been rewritten as to move all else stuff as well  
there).

In addition, moving to the C++ class hierarchy approach is not so  
difficult. I use a template class ref<T> that replaces your "tree"  
class with ref<tree>. You might want to try it, once you have gotten  
your current approach working.

  Hans Aberg