[developers] current activities and code tidying?

Ann Copestake Ann.Copestake at cl.cam.ac.uk
Sun Aug 14 18:11:19 CEST 2005

I propose to do some code tidying to remove the residue of the old LKB
syntax (path notation) assuming nobody is still using it (I will email the
lkb-list to check).  I will retain the option of allowing other syntaxes, just
in case, but will move the files from io-paths into io-general and cut the path
syntax reader.  Say now if you object.

Also - FYI - I am reworking the morphophon / morphosyn interface in order to
allow irregulars - this wasn't difficult in itself but allowing for things with
null affixation meant a rethink.  (The checked in version works but only
because of a bug ...) I believe the revised version will be robust to this and
to cases where affixation decreases the length of the string, but I am testing
on an artifical grammar to try and confirm this.  I will check in the
artificial grammar too when I check in.  I am currently hampered by a slow
connection which has stopped me downloading some of the stuff I need to test to
my laptop.

I am investigating how much of a hassle it'd be to support an optional
specification of the lex ids for irregular forms, so that one could specify
that `hung', for instance, only applied to hang_v_1 and not hang_v_2.  I am
also looking at whether we could replace the *irregulars-only-p* with a more
flexible mechanism that allowed one to specify whether a irregular form blocked
regulars.  Currently, the generator will prefer the irregular form, if one is
specified, the issue is whether the parser will accept e.g., `canned' as a past
tense of `can' or `dreamed' as a past tense of `dream'.  Since I don't want to
have an elaborate syntax for the irregulars file, I think that the right way to
support this is to allow paradigms in the FS.  Obviously one can do this
already, but at the cose of having duplicate rules for the irregular cases -
it would be good to allow this with a single rule.


More information about the developers mailing list