From: Adriano d. S. F. <adr...@uo...> - 2007-05-11 12:12:12
|
Dmitry Yemanov wrote: > Adriano dos Santos Fernandes wrote: > >> I see no problem, but if there is any benefit maybe this could be a job >> to the optimizer. >> > > I'm not sure I understand you. The optimizer cannot generate BLR and I > doubt it should be so smart to replace the parsed nod_equal with > nod_equiv :-) I mean, in no-MATCHING case, fields of the boolean is from a PK, hence they're not NULLable. I don't see difference of nod_equiv and not_equal in this context, maybe some cycles more in nod_equiv. In this case, as the optimizer is smart enough to choose an index when you do "where field = something", it can see "field" is not nullable and transform nod_equiv to nod_equal. Or I'm missing something? Adriano |