Re: I can't think of the subject name (have not words).

"Art S. Kagel" <kagel@bloomberg.net> offerred: +> > An annoyed Informix user wrote: +> Do You see index3 on table1(field1, field2)? There isn't sorting needed. + +WRONG WRONG WRONG! + +> There isn't table reading needed too. index3 contains all information +> queried and in right order. + +ALSO WRONG WRONG WRONG! [ specifics on query and optimization plan deleted ] This is kind of funny, because it illustrates the very reason Informix held off so long in offering "optimizer hints" (what they have implemented as optimizer directives). I worked for Informix for many years in the field, and one of the most common requests from customers was the ability to tell the optimizer how to execute the query, rather than rely on it's cost-based decision making process. The consistent response we used to hear from R&D was "customers always think they know the 'faster way' to execute a query, but often times it is not the faster way; we need to make the optimizer smarter where it is dumb, not give users the ability to override it." Key players in Informix development have departed since that time, and obviously those who were so strongly against "optimizer hints" were no longer around to say no to the addition of this feature. So 7.30 includes the feature, and within a few weeks we see an example of someone complaining about Informix's poor query access plan when the reason it is so poor is that they used optimizer directives incorrectly, i.e. they "told" the optimizer to use a strategy that turns out to be poor. Then the Informix product is blamed for it. Makes me wonder if the previous philosophy was not the wiser one in the long run... Dave -- Dave Kosenko davek@summitdata.com Director of Training Services (732) 469-4070 Summit Data Group (an Informix Authorized Education Center)