Re: DR Grammars

[Tim Josling]

I have emailed Fortes and he told me one of his students is
implementing it into bison. He also sent me the pascal
implementation, and offered to send his PhD paper. I will also
ask if I can cc the emails to the bison and cobolforgcc lists.

See also below

Tim Josling

Hans Aberg wrote:
> 
> Is this an electronic copy?
> 

No it's hardcopy. But I had no trouble finding it at my local
college library.

> Do you know if DR will take any LR(1) grammar?
> 

>From the paper, yes.

> But the parsing is still fully deterministic, right?

Yes

> >At this stage I tend to think that the best solution for the
> >table size in LR(1) is just to put the entries into a hash table.
> >This will be slower but should often be fast enough. Eg if you
> >look at GCC the parse time is negligible as a percentage.
> 
> I think it is would be great to have both, if they are implemented as Bison
> options and it is easy to switch between them. Then one can try them out in
> actual cases to see if there is any performance benefits of the one over
> the other.
> 
> The standard LR(1) algorithm could be implemented with table compression of
> various degrees, so that one can accept larger tables if speed is crucial.
> 
>   Hans Aberg

Agree.