Comments on: SQL Server – COUNT(expression) bug

By: shanghai escort girl

shanghai escort girl — Tue, 09 Aug 2011 07:19:58 +0000

Attractive section of content. I just stumbled upon your site and in accession capital to assert that I get in fact enjoyed account your blog posts. Any way I will be subscribing to your augment and even I achievement you access consistently quickly.

By: vigrx plus result

vigrx plus result — Wed, 11 May 2011 16:15:42 +0000

thanks to the bloggers for sharing true information that is useful and beneficial to readers and other fellow bloggers.

By: Power Balance pas cher

Power Balance pas cher — Mon, 11 Oct 2010 21:18:23 +0000

Hi I love this forumI'm also passionated in extreme sport and strategies to be a lot more efficient...I discovered a piece technology that make me more efficientThanks again for your forumBye power balance pas cher power balance discount cheap powerbalance

By: Phil Brammer

Phil Brammer — Mon, 03 Aug 2009 11:26:36 +0000

Olly,
x.column_id is correct. Because we are left joining to the subquery aliased as ‘x,’ I wanted to show that its value is indeed NULL. If I chose c.column_id instead, we would actually see some values in that column, which is not the intended behavior.

Phil

By: Olly A

Olly A — Mon, 03 Aug 2009 10:11:05 +0000

Wow! That’s really one to look out for. I do think that one of your column headings in the sample data is wrong, though. Shouldn’t ‘x.column_id’ read ‘c.column_id’? Had me scratching my head for a few minutes!

Thanks for alerting us to this – could save a lot of confusion!

Olly

By: Phil Brammer

Phil Brammer — Sun, 02 Aug 2009 02:36:57 +0000

Ed, while you are correct, I think you may be misinterpreting what I’ve written.

The bug as I’ve indicated above, is where COUNT(column) (which only returns a count of NON NULL values) actually returns a value when in reality it should not. If you look at the last resultset I’ve provided, the last column indicates where this bug is present (the cnt_returns_1_y_n column). Clearly you can see that COUNT is returning a 1 where the column it is counting is NULL. THAT is a bug.

Phil

By: Ed

Ed — Sun, 02 Aug 2009 02:29:52 +0000

Actually, from what I’ve been taught in my CURRENT SQL classes (70-441 and 70-442) this is a documented FEATURE, NOT a bug.

It clearly states in the SQL class that I am taking that if WILL NOT get a count of all the records in a table IF YOU DO A COUNT ON A FIELD THAT HAS NULLS BECAUSE A NULL IS JUST THAT – NULL.

As such, the code you listed does exactly what you told it to do – get a count on the field FirstName field WHERE IT HAS A VALUE. NULL doesn’t ‘count’ (forgive the pun) in this case.

For your count to be accurate simply select a field defined as NOT NULL, like the primary key of your data record.