Log In - Home Page

Patch: Using Levenshtein distance to HINT a candidate column name

Edit Patch - Move To Another CommitFest - Delete Patch

CommitFest 2014-12
Topic Server Features
Patch Status Needs Review
Author Peter Geoghegan
Reviewers Robert Haas
Committer Nobody
Close Date (None)
Comments
Patch by pgeoghegan on 2014-03-27 07:12:43 PM: Initial version.
Patch by pgeoghegan on 2014-07-02 10:52:13 PM: New revision addressing concerns around absolute quality of suggestions
Review by michael-kun on 2014-07-17 01:37:15 PM: Reworked the feature while doing a review with 1 patches: 1) levenshtein func move, 2) patch for column hints. Patch 1 could be consider for this CF, patch 2 seems out of scope for now.
Patch by michael-kun on 2014-07-18 07:48:33 AM: Spent more time producing cleaner patch for Levenshtein separation, Patch 2 (column hints) is more or less untouched.
Review by rhaas on 2014-11-07 08:58:27 PM: compiler warnings
Patch by pgeoghegan on 2014-11-10 04:49:06 AM: Revision fixing compiler warnings
Patch by pgeoghegan on 2014-11-13 12:56:38 AM: Broken-out patch that only moves contrib/fuzzystrmatch's Levenshtein distance implementation to core
Comment by pgeoghegan on 2014-11-14 06:00:34 PM: Broken-out Levenshtein-implementation-in-core patch committed.
Patch by pgeoghegan on 2014-11-16 12:39:40 AM: Revision that allows more than one suggestion from a single RTE
Comment by pgeoghegan on 2014-11-26 05:52:50 PM: Final steps?
Patch by pgeoghegan on 2014-12-04 02:22:32 AM: Revision with simplified costing logic and final, absolute suggestion quality test for best match
Patch by pgeoghegan on 2014-12-20 11:19:47 PM: Revision that adds cost of distance of alias to would-be correct alias iff user specified an alias in faulty ad-hoc query (i.e. Tom and Stephen's preferred alias handling).

Add Comment

Please log in to comment on this patch.