Log message #1033350

# At Username Text
# Sep 12th 2009, 16:49 thun_ How do I switch my datasource from within a model for one query then switch back?
# Sep 12th 2009, 16:49 unclezoot with debug = 0, it's now 462ms
# Sep 12th 2009, 16:48 unclezoot thanks for both ideas!
# Sep 12th 2009, 16:48 unclezoot that's where im going wrong, plus async: false is worth investigating
# Sep 12th 2009, 16:48 unclezoot actually i might still have debug =2, i turned it off in the action but not in the ajax action
# Sep 12th 2009, 16:48 BlackIce AD7six: I saw such times before, but that was on a 500mhz P3 with not enough RAM :P
# Sep 12th 2009, 16:47 AD7six unclezoot: that's astronomical
# Sep 12th 2009, 16:47 unclezoot 1.14s
# Sep 12th 2009, 16:47 AD7six unclezoot: what does firebug say is the response time for one of these lookups
# Sep 12th 2009, 16:47 dogmatic69 its just because of something like async: false in the ajax options
# Sep 12th 2009, 16:47 unclezoot and the fields are all indexed etc
# Sep 12th 2009, 16:46 unclezoot yes
# Sep 12th 2009, 16:46 dogmatic69 you can change that
# Sep 12th 2009, 16:46 AD7six unclezoot: are you testing it with debug set to 0
# Sep 12th 2009, 16:46 unclezoot im having a play with a tagging plugin atm which uses remysharp's jquery autocomplete - i noticed that the browser is waiting for the server to return the ajax results, you can't actually type any more letters into your input
# Sep 12th 2009, 16:45 dogmatic69 unclezoot: its really easy to do that in beforeSave and afterDelete
# Sep 12th 2009, 16:45 BlackIce tnx ^^
# Sep 12th 2009, 16:44 unclezoot blackice, good point well made
# Sep 12th 2009, 16:44 unclezoot i like ad7's solution, similar to pseudocoder's solution for building static html in the webroot where possible
# Sep 12th 2009, 16:44 BlackIce unclezoot: yes, but you can prevent that by making/refreshing the cache when a item is added
# Sep 12th 2009, 16:43 unclezoot of course, that's probably irrelevant in this example
# Sep 12th 2009, 16:43 unclezoot i just remember a yahoo article where they said that caching wasnt always the solution, because it presupposes that someone has already made that cache already, in other words, if the user experience is bad on the 1st time they might not come back for the 2nd anyway
# Sep 12th 2009, 16:43 AD7six personally, I treat the entire webroot as a cache folder
# Sep 12th 2009, 16:42 BlackIce besides that it won't be worth it in messyness and maintainability
# Sep 12th 2009, 16:42 BlackIce compared to using a cache instead of database for example
# Sep 12th 2009, 16:42 BlackIce and the init of a class is very quick, you won't win much there
# Sep 12th 2009, 16:41 BlackIce unclezoot: there are opcode cachers like APC for that
# Sep 12th 2009, 16:41 dogmatic69 php has almost no over head... cake is fast
# Sep 12th 2009, 16:41 unclezoot ad7, that sounds like a good compromise
# Sep 12th 2009, 16:41 AD7six unclezoot: a slim controller and debug = 0 means very little overhead
# Sep 12th 2009, 16:40 unclezoot well i was thinking more of the overhead of all the dispatcher initialization etc, just to return a list of db results
# Sep 12th 2009, 16:40 AD7six if you want to be as fast as possible on request 2+
# Sep 12th 2009, 16:40 AD7six unclezoot: make it a get request (/x/whatevertheytyped.json), and when you server it write it to the webroot too webroot/x/whatevertheytyped.json)
# Sep 12th 2009, 16:40 BlackIce what I just said isn't exactly true, but going around cake because your db is slow means your db needs improvement
# Sep 12th 2009, 16:39 BlackIce your problem*
# Sep 12th 2009, 16:39 BlackIce you problem won't be just that autosuggest
# Sep 12th 2009, 16:39 dogmatic69 unclezoot: it will just make a mess
# Sep 12th 2009, 16:39 BlackIce basically, if you have to go around cake you've got a database structure problem
# Sep 12th 2009, 16:39 BlackIce unclezoot: what size of table are we talking about?
# Sep 12th 2009, 16:39 unclezoot yeah i spose, i was just wondering if there's ever any merit in going around cake, specifically for ajax results
# Sep 12th 2009, 16:38 unclezoot which initially could be very small, but potentially could be very big