| |
| |
| |
|
Page: 1 2 3 4 5 6 7 8 9 10 11 12
Comments:
<0> cubert: Data integrity from the point of view of "do all of this, or do none of it". A half done transaction leaves you with failed integrity.
<1> Drk`Angel: do you mean your friend doesn't use cascaded updates/deletes?
<2> i guess i leave all products unharmed and just flag them
<0> Oh god, I don't use cascade updates and deletes either.
<0> I hate side effects. Makes maintenance a bitch.
<2> with mssql on a descent hardware, when can you expect loss in performance from rowcount (if you dont do any fancy optimizing)
<0> I haven't actually seen his code. i don't know exactly what he does or how he does it.
<3> ionix, thats an unanswerable generic question
<2> 10 000 rows? 100 000 rows? 1M rows?
<3> depends on indexes and that manner in which you access the data
<3> the
<0> We just talk theory a lot. He abuses the hell out of temp tables instead of doing m***ive joins.
<2> ok i refrase: what it "a lot of records" in this business?
<3> millions of rows
<2> ok.. so a table with let say, 40 000 rows is nothing to bother over
<3> not at all
<3> better have good indexes though
<2> good.. then i leave them there... my business will not drive those large stores
<3> unless you happen to know Drk`Angel's friend, then you can just call him and have him setup your db without any indexes since they're a waste
<1> I use cascaded updates and deletes and have never had an issue with them. As long as your data is properly normalized and developers aren't doing stupid **** to the data it's fine. ;)
<1> meeting time
<0> sigh.
<0> I dunno why I bother sometimes.
<0> Of course you're right df. Your way is the only way.
<3> oh, i don't mind other ways man
<4> its just theyre wrong
<4> and you're stupid
<4> if you do them
<4> :p
<3> i think its swell when i hear about methods that fly in the face of traditional relational database processes. its 'innovative'.
<3> oh, and yes, they're wrong and you're stupid.
<3> forgot about that
<5> 2 cents: cascades blow
<4> i never use cascades
<4> thge only thing i have against them is accountability
<5> the potential for accident is large
<5> it would be one thing if cascades were enabled, and it was a SQL feature, such as DELETE customer WHERE customerId = 1 WITH(APPLY_CASCADE)
<0> Like I said, I just don't like side effects. Three explicit deletes aren't any slower, and its a lot more obvious what's going on.
<5> I agree. I always take the ordered-teardown approach to deletes involving multiple relations
<6> does my vote count ?
<4> NO
<5> of course
<5> rofl
<6> hehe
<6> Arrakis!
<6> :P
<4> CHOCCIE1!!
<5> brb
<0> btw, I decided to just drop out of that because I didn't see any reason to defend something I don't do and don't know the details about how its done. The jump from "he doesn't do big super joiny selects" to "he doesn't use indexes" just annoyed the crap out of me.
<0> But, I really should have expected it from irc.
<6> well tough luck, its too late now to pull out
<6> you are in
<0> Already did. :)
<6> do you agree dfworking ?
<3> yes, i dont think i'm quite done yet bashing the **** out of your friend and having you be guilty by ***ociation. give me a moment to collect my thoughts.
<6> hehe
<0> Enjoy. I have some work to do tracking down ancient versions of Oracle so I can get exp to work right.
<0> I'll lastlog later, if it'll make you feel better. :)
<3> sure thing man, i have plenty to do just ripping all the joins out of my queries for performance optimization
<4> lmao
<3> BUT I WILL LEAVE THE INDEXES. TELL YOUR BUDDY HE CANNOT HAVE THOSE.
<4> lmao
<7> DELETE FROM webshop_customers_extra LEFT JOIN webshop_customers ON webshop_customers_extra.customer_id = webshop_customers.id WHERE webshop_customers.id is null
<7> whats wrong with that?
<3> which db
<7> mssql
<7> its if i change DELETE FROM, with SELECT
<8> delete one at a time
<8> i mean one table at a time
<3> DELETE w FROM webshop_customers_extra w LEFT JOIN webshop_customers ON w.webshop_customers_extra.customer_id = webshop_customers.id WHERE webshop_customers.id is null
<3> i think he was trying
<3> to delete from webshop_customers_extra
<9> does drhyde_ ever speak anymore?
<3> when there were no matching customers
<7> dfworking: yes
<3> try what i pasted
<7> i was tyring to delete form webshop_customers_extra
<7> okay
<7> Invalid column name 'webshop_customers_extra'.
<3> my bad, didn't replace the table with the alias
<3> DELETE w FROM webshop_customers_extra w LEFT JOIN webshop_customers ON w.customer_id = webshop_customers.id WHERE webshop_customers.id is null
<10> heh
<7> thx htat worked:D
<7> but i didnt learn so much out of ur string
<7> what was so special with the w?
<3> its just how mssql likes its conditional updates and deletes
<8> bah Zico|... refer to "aliases"
<7> okay i will looking at aliases on teh books online
<7> thx ppl
<4> thxpeepz
<11> How would i select lets say top 3 from several diffrent doc types if the types are listed in one column?
<12> group by doc type and order by count
<12> to just get the top 3 you can do SELECT TOP 3
<12> or just fetch 3 rows
<13> is there a way to see all the relationships for a table in a sql query?
<13> im using mssql 2000
<0> http://msdn.microsoft.com/library/default.asp?url=/library/en-us/tsqlref/ts_ia-iz_4pbn.asp
<13> thnx
<14> mssql question: How can I return only the 5th row of a select statement? I know I can set row count or top but that would return all of the top 5. I want just the 5th row
<13> so im tryin to make a script to copy one database to another database is there a really simple way to do that? or is it better to write an app using schema views
<15> scorpions: 2000 or 2005?
<15> SELECT TOP 1 * FROM (SELECT TOP 5 * FROM Table1 ORDER BY ID ASC) AS Top5Table1 ORDER BY ID DESC;
<5> that's it :)
<15> That would work on 2000 and 2005.
<4> hmm
<4> i imported all of these tables and procedures into a sql 2k5 db from 2k
<15> SELECT * FROM (SELECT ROW_NUMBER() OVER (ORDER BY ID) AS row_number, * FROM Table1) AS DerivedTable1 WHERE row_number = 5;
<15> That would work on 2005
<4> and all of the tables and proceudres are username.name
<0> Be a lot easier if mssql just had the rownum psuedocolumn ... and it looks like 2k5 does.
<4> which of course is now changed to schema.name
<15> That should also work on Oracle 10g I believe
<4> and i mae a login which owns the proper schema
<4> but its not finding the objects :S
<15> 2k5 doesn't have a rownum pseudocolumn, but it does implement the ANSI2003 ROW_NUMBER() ranking function
<1> Halo_Four: wouldn't that first one just select the first row from the subselect that has the first 5 rows, basically giving him the first row of the table?
<15> cubert: I reverse the order by
<1> yeah, I see that.. shouldn't the TOP 5 be DESC?
<14> Halo: 2000 - I had a feeling I'd have to do the 2 select but I was hoping for a cleaner solution.
<14> thanks
<0> If you want the 5th from the bottom.
<15> cubert: I grab the top 5 from the top, and then I grab the top 1 from the bottom of the top 5, that would give you row #5
<1> nm, finally wrapped my head around it
<1> :P
<1> I'm slow
<15> While the ROW_NUMBER() implementation looks like a bear given the subselect, it's actually very efficient if the ordered column is an index.
<13> is there an easy way to copy constraints using information_schema? in mssql2000?
<4> :-/
<13> select * from information_schema.CONSTRAINT_TABLE_USAGE;
<15> I found that given a table with 10 million rows, it was only about 2% slower at finding an arbitrary record than if you knew the actual value of the PK of that record.
<13> that lists all my constraints
<1> I'm having a problem where executing a query via ODBC gives an error but running it using the native tools on the DB2 server it runs just fine *sigh*
<1> it's lovely doing debugging on a different platform than you're deploying
<15> What kind of error?
<1> umm.. something about a subselect returning multiple values
<15> hm.
<1> that's what ODBC says, but the DB2 client on the DB2 server has no problem with it
<15> SELECT id, (SELECT blah FROM Table2 WHERE Table2.id = Table1.id) FROM Table1 -- ?
<1> yeah, I'll paste the query
<15> Does DB2 have a method internally to handle heirarchical results?
<1> http://sql-servers.com/nopaste/?show=323
<1> hell if I know
<1> I'm a wannabe developer and an even more wannabe DBA =P
<15> I don't know DB2 but I ***ume there is a way to limit the number of records returned?
<4> there an easy way to change the schema owner of a bunch and a bunch of objects in sql 2k5 ? sp and tables?
<15> Or you can use an aggregate in the subselect, depends on the value that you're looking for
<1> right, but I'm doing multiple subselects and I'm not sure which one is generating the error or why it's only a problem when done via ODBC
Return to
#sql
or
Go to some related
logs:
#beginner
nvidia control panel component to vga hdtv not working
#windowsxp
#computers
dejavu ICOM adventure solution
#windows
#politics
#politics
#windows
#solaris
|
|