| |
| |
| |
|
Page: 1 2 3 4 5 6
Comments:
<0> yeah
<1> There, so I implemented Distinct and now it all works swimmingly :)
<1> If you guys want to see the result of your help, check out www.dailytutor.com/become3.php
<1> (you will have to sign in as someone, or create a new tutor name ... try it, it's fun. I've been working on it for days.)
<2> you developing a online course system ?
<2> an
<3> i have signups
<3> hate
<3> i hate it should be
<0> heh
<0> Parse error: parse error, unexpected $ in /home/content/D/a/i/DailyTutor/html/subjects.php on line 192
<3> uhno
<1> where?
<1> upb: did you go to it straight away? did you go through the signup process?
<0> i tried to sign up
<1> did you go from step 1 and 2?
<0> yes
<0> now i cant even login with null/null
<1> huh
<1> wait, if you went through steps 1 and 2
<1> click here:
<3> :)
<1> www.dailytutor.comm/become3.php
<1> err
<1> http://www.dailytutor.com/become3.php
<1> you shouldn't get any PHP errors. do you?
<0> dno
<0> im going home now, test it yourself
<1> is eveything fine now?
<1> okay i tested it and its great
<1> upb: u just refresh become3.php and it will be ok i thinik
<4> ok i am back...
<4> still stuck on the same problem
<4> i dont think there is a way around my problem without a full table scan
<5> I apologize for reranting, but it boggles my mind how in MSSQL2K that you cannot call GETDATE() or GETUTCDATE() because those functions are indeterministic, and you cannot call sprocs because they can have side effects on the execution context, but you can call xprocs just fine.
<6> lol yea sux
<6> that's why you need to p*** date into the function :(
<4> i basically need to find all the rows in a col that are within X number of permutations to a target string... levenshtein, i know. but there has to be a better way than a full table scan.
<2> well, the full table scan depends on more than the sql statement
<2> at least in my world
<2> if the dbms choose to use a table scan, it's because it is the best way
<4> basically i need every row
<4> isnt taht costly?
<2> you can either try to rewrite the statement to cheat it, or try to set some filters on the result and see what happens
<2> every row is ok, do you need every column in each row ?
<4> no just 2
<4> what if there are 100k rows?
<4> or a million...
<2> that's nothing
<4> that just seems like a bad use of memory.
<5> Back to this, eh?
<4> lol
<2> if you use an IBM mainframe you can talk about cost
<4> yeah
<5> The only way to do it without a brute force table scan is to catalog and index those permutations in a separate table.
<2> how do you select these two columns ?
<5> Oh, and try #mysql.
<2> mysql ?
<4> i did, they are idiots :)
<2> is this mysql ?
<7> haha
<5> Of course they are, but you picked the DB.
<4> no, unfortunately i didnt.
<4> i wish i did.
<4> i would have used psql
<5> Maybe pgres has something built in that you can use. I understand that Oracle has dictionary indexing.
<5> Maybe someone sells or provides a full text engine for MySQL that supports permutations and whatnot, who knows. We don't talk MySQL less to bash it silly.
<4> if i had a table that had the words broken down into "chunks" depending on the size... anything larger than 3 would be broken down into chunks of 3
<4> so each word would have N-2 chunks
<4> indexed the chunks
<7> will that work with your algorithm?
<4> and selected all of the chunks from my target word against the table of chunks
<8> mysql support standards though, cant say that for mssql, it doesnt even know what EXCEPT is
<5> SQL2K5 understands EXCEPT just fine.
<4> tenfour. yes.
<5> Rizard is such a troll.
<4> would it be any faster?
<4> the problem there is i will get alot of garbage data and dupes...
<4> is distinct costly?
<2> how much do you pay for CPU IO gahd ?
<5> Ask #mysql, maybe they know.
<4> lol
<4> these arent mysql specific questions though...
<5> Yes they are. Implementation is always DB specific.
<4> posefant, nothing, but i prefer the app to be responsive.
<2> gahd\neck: :o)
<4> lol
<2> Well, I guess mysql provide a matrix for each operator with the relative cost
<4> in the docs?
<5> http://www.searchinform.com/document-retrieval/mysql-full-text-search.htm
<2> If they do, it should be there or some dude did it for them
<4> full text search doesnt do partials very well from what i can tell
<8> sql2k5 is brand new, no one uses that crap, they know better than to use a new microsoft product
<8> if they dont, they will soon learn the hard way
<2> anyway, lets say you have the word levenshtein, and want all permutations ... are they stored as intersections in your table ?
<9> I plan to migrate in 2007
<5> Riz: Then don't bitch about new standards not being implemented :P
<4> no i dont want all the permutations..
<4> that is a bad idea.
<2> *scrolling*
<5> Only problems I've had with SQL2K5 are with system tables that I attempted to use directly.
<2> ah yeah ...
<4> i think i might have used the wrong word... i want words that are within a treshold of transactions...
<4> i thought permutations was the same...
<9> threshold of transactions?
<4> so like.. THE, if my threshold was 1 i could get THE, TH, TE, HE
<5> You want "levinshtien" to match "levenshtein", right?
<4> or TEH
<4> yes
<5> You don't index words like "the"
<4> exactly
<4> i know
<5> What you want is a decent full text engine which can support this.
<5> I doubt you want to write this crap from scratch.
<4> but i am not going to type out the permutations for the word levenshtein with a threshold of 2
<5> Also, #mysql.
<2> seems more like a decent asm algo to me
<2> select the crap and work on it on the client
<4> that just seems like a bad idea
<4> most of the data is not relavent
<4> i think i like my chunks idea...
<5> This kind of a problem is above the help you'll find on IRC.
<4> but i gotta think it through.. make sure i am not throwing away bad data...
<2> what do you store in this table anyway if I may ask ?
<4> i know.. just wanted to make sure i wasnt over looking something obvious
<4> in this case, business names
<4> its for a yp type directory
<2> why do you need to rearrange the name ?
<9> what mysql version?
<2> k
<4> i dont want to rearrange the names
<9> you want to do phonetic search then, or what?
<4> i want to be able to suggest terms when people dont get any results (maybe from a poorly formed search)
<5> What you need to do is call Google, purchase one of their search appliances, have it catalog your data, and then use it's engine.
<4> i have the metaphones in my table
<2> like www.dictionary.com
<4> lol
<4> nah, i can do this on my own. i dont need a $5k appliance
<2> search for indefinetable and it suggests indefinable
<5> You can't do this on your own, that's why you're spending so much time on IRC asking about it.
<4> i know how to implement this... i am just looking for ways to improve it
<4> Halo_Four: i could write a brute force method in 20 minutes to do this
<6> hmm halo
<5> And brute force is exactly what you don't want.
<6> that profiler thingie isnt helping
<4> yep.
<5> Quant: Sorry, I'll try better next time.
<6> please do
<6> :)
<5> gahd: This is why fulltext engines exist.
Return to
#sql
or
Go to some related
logs:
#vmware
#gentoo
#red
#beginner
#solaris
#online
#solaris
#flash
#politics
#gentoo
|
|