sql like / beseda

Tole da enak resultset, see timing ...

SELECT term FROM tabela FORCE INDEX (id_l2)
WHERE term REGEXP  '^krneki | krneki | krneki$' AND l2 = 'de';
/* Affected rows: 0  Found rows: 23  Warnings: 0  Duration for 2 queries: 8,705 sec. */
RESET QUERY CACHE;
SELECT term FROM tabela FORCE INDEX (id_l2)
WHERE term REGEXP  '((^| )krneki )|( krneki( |$))' AND l2 = 'de';
/* Affected rows: 0  Found rows: 23  Warnings: 0  Duration for 2 queries: 15,055 sec. */

ali se splača razbijat term field na več fieldov s posameznimi besedami zato, da bi izvajal
where f1 = "krneki" or f2 = 'krneki' or ....

Tip hoce presledke okoli. Splitaj pred insertom in naredi to-many relacijo na termine.

FULLTEXT, see timing. Hvala za namig.

RESET QUERY CACHE;
SELECT term FROM tabela
WHERE MATCH(term) AGAINST ('+ krneki+krneki ' IN BOOLEAN MODE)
AND L2 = 'de';
/* Affected rows: 0  Found rows: 23  Warnings: 0  Duration for 2 queries: 0,031 sec. */