| commit | 394eb01c34eb335bb9f1e9f46d8740df769644ef | |
| author | Olly Betts <olly@survex.com> | |
| Mon, 19 Mar 2018 00:01:25 +0000 (19 13:01 +1300) | ||
| committer | Olly Betts <olly@survex.com> | |
| Mon, 19 Mar 2018 00:01:25 +0000 (19 13:01 +1300) | ||
| tree | 45502e8295404b2ead3298f0b12feb04bd7552cb | treesnapshot (tar.gz zip) |
| parent | b01e16cf3213836f6a578aa5bdc630c1ba178f48 | commitdiff |
[honey] Don't store "plateaued" wdf values
If all the wdf values for a term are the same (with the possible
exception of the first) then we can detect this and don't need to
actually store them. This reduces the size of the postlist table
in a sample database of wikipedia pages by 1.7%.
If all the wdf values for a term are the same (with the possible
exception of the first) then we can detect this and don't need to
actually store them. This reduces the size of the postlist table
in a sample database of wikipedia pages by 1.7%.