|Convincing an LR parser to emit fragments? firstname.lastname@example.org (1992-05-26)|
|Re: Convincing an LR parser to emit fragments? email@example.com (1992-05-27)|
|From:||firstname.lastname@example.org (Stephen P Spackman)|
|Organization:||University of Chicago CILS|
|Date:||Wed, 27 May 1992 03:51:56 GMT|
|My query concerns the use of LR parsers for dealing with undergeneration.
|A chart parser will produce a chart showing the set of partial derivation
|sequences that it has constructed. Would it be possible to convince an LR
|parser to do likewise?
Perhaps it's important to ask if this is the *expected* case or not.
Certainly that information is present in the parser state, but I think
it's fair to say that the LR technique gets its leverage by encoding it as
much as possible.
(Another question that MIGHT be worth considering, depending on your
application, is if the partial parses are the best possible output, or
if it's just the thing that the chart parser makes easy; for instance,
I seem to remember reading of an LR-based minimal completion algorithm
recently - perhaps in SIGPLAN Notices? - which struck me as a useful
tool in feedback situations.)
stephen p spackman Center for Information and Language Studies
email@example.com University of Chicago
Return to the
Search the comp.compilers archives again.