|
|
Posted by Hugo Kornelis on 06/27/07 17:39
On Tue, 26 Jun 2007 17:27:42 -0700, --CELKO-- wrote:
> The illegal
>syntax you used can do multiple updates on each row; talk to an old
>Sybase programmer about this problem.
Hi Joe,
It's not illegal. It's non-standard, but in SQL Server (check the name
of this group!!), it's legal and documented. And in many cases much
faster than the ANSI-standard equivalent - an ommission I keep kicking
the optimizer team for until they finallly get it right, but until that
day we developers have to code our way around it.
>UPDATE A SET field1
> = (SELECT C.field1 FROM B, C
> WHERE A.guid1 = B.guid1
> AND B.guid2 = C.guid2 AND C.somekind_type = 1
> AND @my_date BETWEEN B.start_date AND B.end_date);
Oh boy, I really hope that Richard runs a quick test or two (oh wait,
one should be enough with such glaring errors) before deploying this
"alternative" in production.
I think that this snippet of code is a good contender for not only the
worst code ever posted to usenet, but also for the worst formatting. If
I were a book author, I'd now have to ask you for permission to include
it as a sample of bad query writing (though the reader probably wouldn't
believe it -hah! as if I even *could* come up with something this bad).
--
Hugo Kornelis, SQL Server MVP
My SQL Server blog: http://sqlblog.com/blogs/hugo_kornelis
Navigation:
[Reply to this message]
|