Re: Machine grammar and elidables

[cowan (John Cowan)]

David Bowen writes:
>    I was looking at this issue a few weeks ago.  I concluded that we could
> eliminate "cu" with only a minor change in the grammar, without introducing
> any ambiguities.

In the grammar the parser actually implements, "cu" is no longer an elidable
in the formal sense -- it is actually optional, i.e. there are separate
rules for cu-present and cu-omitted.  That's why your test worked.
"fa'o" is also now treated specially, not as a regular elidable.

As for the others, "ku'o" is the only one known to be always elidable
if nothing else is elided.  All the others are needed at least sometimes,
and we keep "ku'o" for symmetry and because it sometimes makes it possible
to elide more than one consecutive terminator by inserting it.

We didn't define the elidables for fun.  All of them, modulo the three
special cases here, are required to keep conflicts out of the grammar.
Many were put in specifically because the presence of conflicts was
reported during machine-grammar development.

-- 
cowan@snark.thyrsus.com		...!uunet!cbmvax!snark!cowan
		e'osai ko sarji la lojban