Re: A tricky grammar

> They also observe that GLR is harder to implement than The.

Scott, Johnstone, and Hussain published “Tomita-Style Generalised LR
Parsers” (2000)[*] which discusses some weaknesses in Tomita’s original
algorithms and goes on to describe a “Generalised reduction modified LR
parser” which they express in about a page of code. (A page of somewhat
dense code festooned with symbols, but that’s not wholly unlike the page
or so I implemented for my current parser.)

The prospect that a DFA based parser might create fewer objects and
consequently put less GC pressure on the JVM does inspire me to consider
attempting it. But not *this* weekend.

> Earley crucial point, is probably one Tom made already: new
> implementations of ixml are always welcome!

Indeed!

                                        Be seeing you,
                                          norm

[*] You will have no trouble finding it, though I cannot work out what a
canonical location might be, so I’m leaving it to your search engine of
choice.

--
Norm Tovey-Walsh
Saxonica