Re: A tricky grammar

> In wisps-001-020, sample grammars 5 and 6 both have two-symbol loops.  I
> think NineML passed those tests, but perhaps you have touched the code
> since then.

I don’t think so. But I was hand waving a little bit. I think that two
conditions had to be true simultaneously for the bug to arise: first,
the loop had to involve more than once symbol and, second, one of the
nodes in the shared packed parse forest involved in the loop had to have
a particular configuration of child pointers. (Specifically, or at
least, it had to have two children.)

> And until I get around to moving the wisps tests into the
> ixml/tests directory, I suspect they won't be part of your routine
> testing.

True. That testing isn’t part of the mandatory, automated regression
testing required to make a new release.

                                        Be seeing you,
                                          norm

--
Norm Tovey-Walsh
Saxonica