Tested; searches work well, but we're back to the old problem of metabib.title_field_entry rows being duplicated:
SELECT * FROM metabib.title_field_entry WHERE length(value) < 20 AND field = 6 ORDER BY source;
id | source | field | value | index_vector -----+--------+-------+---------------------+--------------------------------------------------------------- 3 | 2 | 6 | Le concerto / | 'concerto':2A,4C 'le':1A,3C 4 | 2 | 6 | Le concerto / | 'concerto':2A,4C 'le':1A,3C 64 | 22 | 6 | The concerto / | 'concerto':2A,4C 'the':1A,3C 65 | 22 | 6 | The concerto / | 'concerto':2A,4C 'the':1A,3C 207 | 51 | 6 | Cello concerto. | 'cello':1A,3C 'concerto':2A,4C 208 | 51 | 6 | Cello concerto. | 'cello':1A,3C 'concerto':2A,4C 219 | 53 | 6 | 6 Double concertos. | '6':1A,4C 'concerto':6C 'concertos':3A 'doubl':5C 'double':2A 220 | 53 | 6 | 6 Double concertos. | '6':1A,4C 'concerto':6C 'concertos':3A 'doubl':5C 'double':2A
Oh, wait - that's bug 1032208 which never ended up getting resolved (but you would think that trimming out a large percentage of rows should have some effect on search performance). So I guess this is back to normal, then. Excellent!
Tested; searches work well, but we're back to the old problem of metabib. title_field_ entry rows being duplicated:
SELECT * FROM metabib. title_field_ entry WHERE length(value) < 20 AND field = 6 ORDER BY source;
id | source | field | value | index_vector ------- +------ -+----- ------- ------- --+---- ------- ------- ------- ------- ------- ------- ------- ------- ---
-----+-
3 | 2 | 6 | Le concerto / | 'concerto':2A,4C 'le':1A,3C
4 | 2 | 6 | Le concerto / | 'concerto':2A,4C 'le':1A,3C
64 | 22 | 6 | The concerto / | 'concerto':2A,4C 'the':1A,3C
65 | 22 | 6 | The concerto / | 'concerto':2A,4C 'the':1A,3C
207 | 51 | 6 | Cello concerto. | 'cello':1A,3C 'concerto':2A,4C
208 | 51 | 6 | Cello concerto. | 'cello':1A,3C 'concerto':2A,4C
219 | 53 | 6 | 6 Double concertos. | '6':1A,4C 'concerto':6C 'concertos':3A 'doubl':5C 'double':2A
220 | 53 | 6 | 6 Double concertos. | '6':1A,4C 'concerto':6C 'concertos':3A 'doubl':5C 'double':2A
Oh, wait - that's bug 1032208 which never ended up getting resolved (but you would think that trimming out a large percentage of rows should have some effect on search performance). So I guess this is back to normal, then. Excellent!