IRC Logs

Network: freenode Channel: #csharp

	Enter your search terms Submit search form
	Web www.irclog.org

[02:30]<kjr>`m0: I'm looking at an EXPLAIN SELECT
[02:30]<mjzymygo>Kog: port everything to MSSQL/Oracle and MYSQL.Dispose()
[02:30]<mjzymygo>;)
[02:30]<kjr>`m0: looks like it's not respecting indices
[02:31]<`d0>hmm
[02:32]<`d0>which type of indices?
[02:32]<`d0>it wont tell you much
[02:34]<`d0>it shouldn't refuse indexes. What type of indexes did you place?
[02:34]<kjr>it's mysql... an INDEX is an INDEX type
[02:34]<`d0>some storage engines wont allow you to use some of the index types
[02:35]<kjr>no, something is wrong here
[02:35]<kjr>I'm gonna check over all the joins again
[02:35]<kjr>look for missed indices
[02:35]<kjr>I'm also troubled by the fact that some had a cardinality of NULL
[02:35]<kjr>which shouldn't be possible except (maybe) in an empty table
[02:35]<kjr>I might have gotten some corruption when I had to force-kill the SQLD
[02:35]<`d0>what type of joins are you using?
[02:35]<kjr>LEFT
[02:36]<`d0>that produces cardinality of NULL...
[02:36]<kjr>huh? no
[02:36]<`d0>if some rows are not included
[02:36]<kjr>the indices have a cardinality of NULL
[02:36]<`d0>erm matched
[02:36]<kjr>the indices shouldn't ever have a cardinality of NULL if there's data in the attribute
[02:36]<kjr>ever.
[02:36]<`d0>so there is data?
[02:37]<`d0>thats strange :)
[02:37]<kjr>yes, very
[02:37]<sdnffdxdlld>woo its gonna be a nice 24/23 degrees in the dam when I go :)
[02:37]<izit>:D
[02:37]<izit>smellyhippy: it works perfectly!
[02:37]<sdnffdxdlld>of course it does 8)
[02:37]<kjr>oh hell... more of the indices are null
[02:38]<`d0>hmm Kog innoDB?
[02:38]<kjr>MyISAM
[02:38]<mjzymygo>MyISAM.Dispose()
[02:38]<`d0>how are you referencing tables?
[02:38]<`d0>lol Moridin8
[02:38]<mjzymygo>MYSQL.Dispose()
[02:39]<`d0>private MYSQL.Show()
[02:39]<kjr>`m0: explicit LEFT JOIN
[02:40]<kjr>I hope this is just phpMyAdmin hallucinating
[02:40]<`d0>I believe indeces should be null if you don't have a unique/foreign key
[02:40]<kjr>no
[02:40]<`d0>| 1 | SIMPLE | category | ALL | NULL | NULL | NULL | NULL |
[02:41]<`d0>i dunno
[02:42]<`d0>never had that problem
[02:42]<kjr>I'm asking in #mysql... but that channel can be sometimes unhelpful
[02:42]<kjr>`m0: part of the problem is that the database is outsourced
[02:43]<`d0>wait till midnight till the aussies are awake
[02:43]<sdnffdxdlld>Moridin8: your very good at not noticing new tabs appering in mirc arnt ya ;)
[02:45]<mjzymygo>sorry dude... finishing up some work...
[02:53]<mjzymygo>gotta go sleep... be cool peeps
[02:57]<drdyrrr>Heh, just did some benchmarks with .nettier orm
[02:57]<drdyrrr>my god is it slow and chunky.
[02:57]<sdnffdxdlld>heya dude
[02:57]<sdnffdxdlld>your up late
[02:57]<izit>Going to slepp now. Thanks for the help
[02:58]<sdnffdxdlld>np bud :)
[02:58]<txnjzd>damieng: is it fast enough?
[02:59]<drdyrrr>simple table with 6 fields and 30,000 records.
[02:59]<drdyrrr>only field nullable.
[03:00]<drdyrrr>do a .GetAll - the same as a select *....
[03:00]<drdyrrr>5.7 seconds, 42MB
[03:00]<drdyrrr>compared with my existing homebrew orm;
[03:00]<drdyrrr>0.195 seconds, 4MB
[03:00]<txnjzd>are they doing the same thing?
[03:01]<drdyrrr>they both give me a business object with change notification
[03:01]<maxgg_>did someone just say homebrew orm?
[03:01]<drdyrrr>and a data mapper class with a level of abstraction for an underlying provider
[03:01]<txnjzd>do datasets not do all this anyway?
[03:01]<drdyrrr>and they are both code generation in c#
[03:01]<drdyrrr>err, no.
[03:01]<maxgg_>holy batman architectual astronaut + not invented here + programmer turned architect!!!
[03:02]<drdyrrr>dchen_?
[03:02]<drdyrrr>i was expecting it to be a bit slower... but not that bad.
[03:02]<drdyrrr>i'd hate to think what these reflection based orm's are like
[03:02]<sdnffdxdlld>thats not reflection based? :O
[03:02]<txnjzd>I use a reflection based orm
[03:03]<txnjzd>I've not benchmarked it, but it's fast enough
[03:03]<drdyrrr>well the 30,000 records is from a tiny small test system
[03:03]<drdyrrr>reality could well be 100 to 1000 times that
[03:03]<sdnffdxdlld>damieng, does that time include the roudn trip to the db? or jus tthe obejct creation?
[03:03]<txnjzd>damieng: and you want to get all of them?
[03:03]<ezyrr>anyone here certified?
[03:04]<drdyrrr>i took three benchmarks.
[03:04]<drdyrrr>first one was init time including getting a record count to make sure the db was open.
[03:04]<drdyrrr>second one was to getall
[03:04]<drdyrrr>third one was another getall to see if it caches it
[03:04]<txnjzd>does it?
[03:04]<maxgg_>if you guys keep talking about orms
[03:04]<drdyrrr>well it uses some of the enterprise library to do db caching
[03:05]<drdyrrr>but i've not delved into every bit of code it produces
[03:06]<maxgg_>Erica: i am certified
[03:06]<maxgg_>i have gone loony
[03:07]<drdyrrr>Bench figures for .nettier were; init 265ms, 129 KB; first run 5671ms 49,563 KB; second run 5828ms 49,838 KB
[03:07]<sdnffdxdlld>there had to be someone who was gonna be that cheesy :P
[03:08]<drdyrrr>matching figures for my own is 578ms 121 KB, 109ms 4,624 KB, 94ms -426 KB
[03:09]<txnjzd>second run slower?
[03:09]<txnjzd>effective cache :-)

Back to #csharp
Or go to some related logs:
#gentoo
#netbsd
#gentoo
#suse
#php