Re: silvafind
Marc Petitmermet <[email protected]>
| Newsgroups | gmane.comp.web.zope.silva.devel |
|---|---|
| Message-ID | <[email protected]> |
On 16. Mrz 2007, at 11:10 Uhr, eric casteleijn wrote: >> what about the performance impact because of this additional loop? >> i guess (but i don't have any hard figures) that overall the >> search results are presented faster because the second batch loop >> is smaller and there are less conditions to be evaluated and less >> translations to be made. > > > The performance impact is not so much in the loop, as in that with > this patch, every single object in the result set, not just in the > batch, and that for every page view. oh, well. i have to agree, it was late and i did not think about this. we could at least make sure that "results python:[item for item in resultparts[0] if model.isViewableForUser(item)]" is only evaluated once under the condition="not:batch" or something similar... and if you don't allow empty string searches and don't allow common words (i mysql these words are called stopwords [1]) the impact should be acceptable because the result sets are probably not huge... regards, marc [1] http://dev.mysql.com/doc/refman/5.0/en/fulltext-stopwords.html