| |
| |
| |
|
Page: 1 2 3 4 5 6
Comments:
<0> especially when someone else has the regular expression all ready for ya
<1> the regex should make the where statement fairly simple
<0> it's just gonna be slow, try a text search like verity if you need the speed
<0> well yeah, thats sorta what I was sayin
<1> but.. do i save anything over just a brute force compare EVERYTHING in the column?
<0> doubt it, like I said, it really sounds like a job for a text search engine like verity (and I mention that because it's the only one I've used)
<1> text search?
<1> i thought regex was supposed to be really fast....
<0> yeah, they ***ume huge flatfiles and build 'collections' based on what you're looking for
<1> just not in this case?
<0> anytime regular expressions are used slows things down, doesn't matter where, perl, php, sh, cold fusion
<0> thats been my experience
<0> now WOW they have then in SQL servers
<0> woopie
<1> wierd.
<1> so what use are they?
<1> dinosaurs?
<0> great use to people who know regular expressions and can't do things any other way
<2> nah
<2> they're great if you know 'em at all
<0> it is a great great tool for finding patterns in flatfiles, but it has never been known for its speed, essentially a sysadmin's tool
<2> it's essentially for getting things done fast in human time, not neccesarily cpu time
<2> which imho is of more value
<0> but most cases where they are used they could have just made a decently thought out where statement, created an index just for it and it would blaze
<2> true for repeated use cases ;-)
<0> well thats fine, for sure, but slow statements in a relation database can hurt your concurrency
<2> I mostly throw in regexes when I'm not exactly sure what I'm looking after, but can't be bogged down w/ false positives
<0> if it's a one off, or a report go for it
<3> yay, just 4.5million rows to go....
<1> not a one off...
<0> who cares if the report takes 10sec or 45
<1> no.. this cant take any human time at all
<0> but on a busy db that 10sec query would create havoc with the rest of the requests
<1> its used to return search results
<4> this is a really stupid discussion
<0> who asked ya
<0> ...and I really don't even care who ya are
<4> i'm your boss. now get to work!
<0> funny, my boss reports to me, so YOU get to work
<4> but my job is to boss you around
<0> can't refute that...you win, laterz
<1> why is it a stupid discussion?
<0> hahah...question in #MSSQL: [12:17] <TJojje> When does the logfiles truncate?
<3> what's so haha?
<0> SQL server, filling up logfiles since 1987
<3> isn't that teh ****
<5> sql server for windows 1.0?
<0> I was exhaggerating
<0> microsoft bought thier codebase for SQL server from Sybase, don't remeber when
<5> go figure :P
<0> back in 97, the two were almost indistinguisable
<5> 1993 was 4.21 for NT
<0> only sybase ran on those huge sun boxes with fast busses
<5> in 89 they were porting it to os2
<0> wow: I have never seen this one before: Server: Msg 8650, Level 13, State 127, Line 1
<0> Intra-query parallelism caused your server command (process ID #70) to deadlock. Rerun the query without intra-query parallelism by using the query hint option (maxdop 1).
<0> first time SQL server ever told me to do a hint
<0> oracle 8i's optimizer was dumb as a post (meaning you always had to hint indexes)
<6> haha so database mirroring is evaluation only?
<6> it wasn't posed that way during beta.. funny.
<0> heh...there's a good reason for that, I'm sure
<0> I remeber trying to get NT4's mirroring to work and it cuased more problems than it solved, certainly...and it was for a SQL server too... eventually I found it to be a hardware problem (probably why I hate SCSI drives) with the drive... But it was a thermal problem so it would work sometimes, but when it broke, the mirroring didn't detect it and keep the server up... thats when I decided to hell with it
<7> SQL 2000: I have a select statement in a stored proc which is performing a "clustered index scan". When I pull it out of the proc and execute it in Query Analyzer, it uses an "Index Seek" on the appropriately designed index. Has anyone else seen this kind of behavior?
<8> We've seen just about everything at one point or other.
<7> heh..good point
<8> does the procedure have parameters?
<7> It does, and in QA, I've Declared the variables and ***igned the exact same values that are being supplied to the stored proc
<8> doesn't matter; there was probably a poor plan in the cache from a prior execution, and it was reused
<7> I thought about that, so I've issued a "DBCC FREEPROCCACHE", but it still didn't change the execution plan
<9> EXEC sp_recompile 'MySproc'
<7> No effect
<8> when you run it in QA, are you running the SELECT with variables in place or with constants?
<7> Variables
<7> Just like it looks in the proc
<4> how do you guys deal with create scripts & source control? do you find that merges / conflicts are a problem?
<4> or do you require separate create scripts for each object?
<0> I'd probably try hinting the index in the sproc, it's gotta be better than the clustered scan
<8> that's weak, though
<0> I must say though I've never seen the optimiser be that dumb
<7> HA! Figured it out.... I just did an "UPDATE STATISTICS WITH FULLSCAN"...now it's using the right index. CRAZY!
<7> Me either
<0> good one, I'll remeber that
<7> What's odd is I did the update stats WITHOUT the FULLSCAN before, but that didn't do anything
<7> Thx for the help, guys
<0> analyser shows the stats problems though, right?
<8> update statistics probably was not what helped you
<7> no, it didn'
<8> keep running the test
<7> didn't
<0> huh, well thats good t know
<8> What I'm saying is that UPDATE STATISTICS, with our without FULLSCAN, will flush plans from the cache
<8> you may be seeing a side effect, rather than a solution, and you must continue testing
<0> I see
<0> I gotcha
<7> I've tried everything, though. FREEPROCCACHE; sp_recompile; tried creating another index...and even, previously, updated the stats.
<7> I have heard, from Microsoft, that every so often doing an Update Stats with FULLSCAN is highly recommended.
<3> tenfour whaddya mean, why would having create scripts in source control be a problem
<7> I even created a secondary proc, all the same, but with a different name to test...that didn't work either.
<8> POWER2112, what I'm saying is you cannot simply say "I did A. Next I observed B. Therefore, A causes B." That is engaging in a logical fallacy, even if it proves to be true later on.
<4> if they are generated from enterprise manager they will be in the complete wrong order
<4> making diff/merge impossible
<7> Gamble: Good point...noted
<8> tenfour, purchase a change management tool.
<3> tenfour indeed, maybe you need a tool that does a better job of ordering the ddl generation
<4> i would love something that just takes a create script & orders them in a deterministic way
<3> or maybe something that can sort it for you prior to checkin
<4> yea
<8> you basically want to compare the scripts on DatabaseA with those on DatabaseB ?
<4> nah just maintain schema changes when multiple devs are working on the db
<8> uh, basically same thing, right?
<4> yes but i don't think this requires a system change mgmt solution
<4> it's really just about scripts, not the system
<10> shouldnt developers be working with code that is stored in a change management system outside the database?
<0> heh..tenfour, I made a Cold Fusion Script that get details from the system tables and makes all the create scripts of which I store in SourceSafe
<8> so if dev's change 20 tables, 40 stored procedures, and 15 views on two systems, you're gonna create the "synch" scripts by hand?
<0> but thats a custom solution
<8> Guest, that sounds like DROP/CREATE
<0> well it is
<3> isn't there a dba tool that does ddl diffs and such automatically....
<8> there are tools for this already; Embarcadero and Red-Gate come to mind.
<0> DBArtisan is awesome...
<3> ah, I think that's what I was thinking of
<10> toad does it ok for oracle
<8> Doing those tasks by hand is like making your own rake to go clean up leaves
<3> dbartisan
<0> especially if you're a multi-platform house
<10> doing it by hand is best in my opinion
<8> ;)
<4> if my rake cost $5000 i'd make my own
<3> doing it by hand is all a pain in the ***
<4> huhu
<10> agreed tenfour :P
<3> had to restart my table import :(
<0> do it by hand a few times...by the third time around you'll be writing a script
<4> anyway this is a small project; we're doing fine the way it is (by hand)
<8> tenfour, even at the low end of the scale, if you waste 100 hours doing this over a project, you already wasted the 5 grand
<10> I have a large project and all code is managed "by hand"
<8> good for you
<10> you cant just trust some tool to extract all your ddl and ordered properly pulled from a dev database
<4> 100 hours managing sql script checkins is crazy
<0> yes, and his wife loves him for it
<4> the only thing barring the scm from doing it automatically is simply the order in which EM generates the scripts
<0> always home on time, never works weekends
<8> CoJoNEss, I couldn't agree more; I verify the scripts, and generally I do them in small batches, rather than one-shot the deal.
<8> But no way can I manage it in a database, for example, that has 1500 tables and 3000 stored procedures.
<10> which should all have create scripts that should be made during development
<10> not after the fact
<3> 1500 tables, eeek
<3> wtf does one do with 1500 tables...
<10> mostly lookup tables in my case :P
<0> ahahaha
Return to
#sql
or
Go to some related
logs:
how to close xemacs
3ds pbrt
pulheteiros
#computers
#stocks
galaxyweb MUD
#windows
yourexgirlfriends torrent
#hardware
#beginner
|
|