Re: Bison and LR(1) Parsers

[Hans Aberg]

At 07:12 +1000 2001/07/18, Tim Josling wrote:
>I want to finish the current phase of my compiler before doing
>anything to bison. This will be in a couple of months eg
>September/October.

I think that the plans to add new algorithms to Bison are in no hurry.

> I am very confident that full LR(1) would
>satisfy my needs. The Fortest approach does implement full LR(1)
>function, though in a new way. The question is whether the Fortes
>approach is best, or whether standard LR(1) would be best.

I am not sure any of them will prove to be _the_ best. At Berkeley, they
work on what might be viewed as the successor of Bison's semantic feature,
based on algorithms such as GLR. Different algorithms might work in
different contexts.

>If someone implements the Fortes approach in a reasonable time, I
>would use that, otherwise I would propose to write vanilla LR(1)
>functionality.

My hunch is that you should not tie your efforts to any eventual
developments of the DR algorithm.

Possibly you should instead coordinate your efforts with the Berkeley
group, as they are developing their own version of Bison/Yacc. You might
try contacting Susan L Graham <address@hidden>, which once upon the
time was the thesis advisor of Paul Corbett who wrote Bison.

  Hans Aberg