December 14, 2011 at 9:45 pm
bitbucket-25253 (12/14/2011)
Jeff Moden (12/14/2011)
BWAA-HAAA!!! Judging by the fact that more people got the wrong answer than the correct one, I'd say you just taught a whole bunch of people a lesson, Ron. Well done! 🙂Coming from some one like yourself and your anti-RBAR, quirky update and so many other articles which teach so many so much (and I am one of those you have taught) I thank you for the compliment
I wish I said that.
December 15, 2011 at 4:57 am
Revenant (12/14/2011)
bitbucket-25253 (12/14/2011)
Jeff Moden (12/14/2011)
BWAA-HAAA!!! Judging by the fact that more people got the wrong answer than the correct one, I'd say you just taught a whole bunch of people a lesson, Ron. Well done! 🙂Coming from some one like yourself and your anti-RBAR, quirky update and so many other articles which teach so many so much (and I am one of those you have taught) I thank you for the compliment
I wish I said that.
Thank you both for the wonderful feedback but this one is all Ron and a very well chosen/written question. When I first looked at the question, I thought 98% would get it right. Amazing the way it turned out.
Ron, there's a whole lot of information and misinformation out there about sp_* especially since code no longer really lives in master but in the mostly-hidden "resources" DB. This would make a wonderful article even if it were only a short "Spackle" article.
--Jeff Moden
Change is inevitable... Change for the better is not.
December 15, 2011 at 6:35 am
Thanks Ron. This is good to know.
Jason...AKA CirqueDeSQLeil
_______________________________________________
I have given a name to my pain...MCM SQL Server, MVP
SQL RNNR
Posting Performance Based Questions - Gail Shaw[/url]
Learn Extended Events
December 15, 2011 at 11:32 am
Hugo Kornelis (12/14/2011)
This actually makes the behaviour even more weird, since (according to the documentation for SQL Server 2005 - I see the relevant comments have been removed from the SQL 2008 docs, which is even more disturbing) stored procedures with a name starting with sp_ are searched in master first, user DB second.
Yes, weird as heck. But now I'm really confused about the precedence of similarly-named procedures. Running this on SQL Express 2005, the procedure in user db "Experiment" is executed while connected to that db, while the master db version is run while in master or any other db (like model).
use master
go
If object_ID('sp_myproc') is not null
drop procedure sp_myproc
go
Create procedure sp_myproc --create in master
as Select 'sp in master db.' as ProcOutput
go
Exec sp_myproc
go
use Experiment
go
If object_ID('sp_myproc') is not null
drop procedure sp_myproc
go
Create procedure sp_myproc --create in experiment
as Select 'sp in Experiment db.' as ProcOutput
go
Exec sp_myproc
go
use master
go
Exec sp_myproc
go
use model
go
Exec sp_myproc
go
edit to correct mistaken comments in the sql code.
December 15, 2011 at 12:07 pm
After finding the BOL doc that warns against naming a procedure with an "sp_" prefix, I wonder if the precedence rules are controlled by another factor besides the prefix and database. The doc does say that a procedure whose name matches a system procedure would be overridden, but perhaps we're assuming too much that if we create our own "sp_xxxxxx" in the master db, that it would have the same preeminent status.
If your application uses nonschema qualified name references and your own procedure name conflicts with a system procedure name, your application will break because the name binds to the system procedure, not your own.
Is there another factor besides the sp_ prefix and being in the master db that controls this? Perhaps the key to this is in that "resources" db that Jeff mentions.
December 15, 2011 at 12:52 pm
john.arnott (12/15/2011)
Is there another factor besides the sp_ prefix and being in the master db that controls this? Perhaps the key to this is in that "resources" db that Jeff mentions.
Yes, there is some other factor, but I haven't a clue what it is. In fact many stored procs whose name begins sp_ and which are supplied by MS as part of SQLServer, referred to in BoL as system procs, and listed by SMSS object explorer as system stored procs turn out not to be system stored procs at all as was pointed out to me recently by one of the SQLServerCenter gurus (can't remember who).
Tom
December 15, 2011 at 1:23 pm
Reposting these links from earlier, for those looking for the details:
http://blogs.msdn.com/b/sqlprogrammability/archive/2006/04/03/567643.aspx
http://blogs.msdn.com/b/sqlprogrammability/archive/2006/04/03/567648.aspx
http://blogs.msdn.com/b/sqlprogrammability/archive/2006/04/03/567660.aspx
http://blogs.msdn.com/b/sqlprogrammability/archive/2006/04/03/567663.aspx
December 16, 2011 at 12:57 pm
Did not know that. Thanks!
Chris Umbaugh
Data Warehouse / Business Intelligence Consultant
twitter @ToledoSQL
December 20, 2011 at 5:33 am
Good question, thanks for submitting. Not everyone is aware of this functionality and I've ran into it on several occasions.
http://brittcluff.blogspot.com/
December 20, 2011 at 5:22 pm
Toreador (12/14/2011)
Nice question, but not sure about the explanation. Surely the crucial point is that object_id('sp_mytable') is null, even though a select against that table recognises it (and a drop would drop it)?
Well, I'd say there were two crucial points, (1) that you can refer to a table in master without a fully qualified name while in another database, and (2) that object_id requires the fully qualifed name if the object is not in the current database.
I think the answer would be improved by pointing out the inconsistent behavior, as it is, without an explanation, the 2nd drop serves as a guard to ensure the correct response, but there's no explanation as to why it's necessary (or the consequences of using it if your QOD table happens to already contain a sp_mytable).
December 22, 2011 at 6:18 am
Nice!
I forgot that.
December 22, 2011 at 1:50 pm
Good Question...
Blog -- LearnSQLWithBru
Join on Facebook Page Facebook.comLearnSQLWithBru
Twitter -- BruMedishetty
December 26, 2011 at 9:50 pm
December 27, 2011 at 11:48 am
Great question and great information. Always good to learn something new or re-learn the forgotten!
February 8, 2012 at 8:59 am
Great question... Is that mean when any object with SP_ prefix used in combination with if (as in this case) would be searched in the current DB first and then will go for master DB? Even if so, won't the if statement given in the question do a second search in the master DB automatically? I am bit confused... :rolleyes:
Viewing 15 posts - 31 through 45 (of 52 total)
You must be logged in to reply to this topic. Login to reply