Comments on: How to Break a MySQL Left Join

By: MySQLuser

MySQLuser — Wed, 10 Feb 2010 06:24:05 +0000

I have found interesting post about Conditional Joins at
http://www.mysqldiary.com/conditional-joins-in-...
Nice example with good explain. Helped me a lot.

By: Aaron

Aaron — Sat, 17 Oct 2009 02:19:03 +0000

Great article. It pointed me in the right direction. Thanks for doing this, these kind of blogs save a lot of headache!

By: Greg Allard

Greg Allard — Thu, 08 Jan 2009 20:43:32 +0000

I'm glad that this has been able to help other people out

By: Shane

Shane — Thu, 08 Jan 2009 20:33:51 +0000

You are awesome, stumbled upon you today after spending hours and hours stumbling about frustrated.

By: Prashant Deepak Vedpathak

Prashant Deepak Vedpathak — Tue, 05 Aug 2008 08:08:23 +0000

Thanks Greg and Sean.

It helped me to reduce me number of sec to execute my query by using the left join and coalesce concept.

By: Prashant Deepak Vedpathak

Prashant Deepak Vedpathak — Tue, 05 Aug 2008 07:08:23 +0000

Thanks Greg and Sean.

It helped me to reduce me number of sec to execute my query by using the left join and coalesce concept.

By: Greg Allard

Greg Allard — Thu, 08 May 2008 16:10:03 +0000

Ah it would be COALESCE(log_deleted, 0) = 0 for MySQL

By: Greg Allard

Greg Allard — Thu, 08 May 2008 16:05:50 +0000

Thanks for the responses. I'll have to test out the performance of using OR in MySQL and see if it has similar performance loss.

It looks like that would only be an alternative for MSSQL since the ISNULL function only has one parameter for MySQL. I'll keep an eye out for something similar in MySQL.

By: Sean

Sean — Thu, 08 May 2008 15:42:43 +0000

Oh, it might be better to write the query like this:
SELECT task_text, count(tasks.task_id) AS log_count,
sum(log_length) AS total_time
FROM tasks LEFT JOIN logs ON tasks.task_id = logs.task_id
WHERE task_deleted = 0
AND isnull(log_deleted, 0) = 0
GROUP BY logs.task_id;

By: Sean

Sean — Thu, 08 May 2008 15:41:32 +0000

My experience with this sort of join, in MSSQL is that putting an or on the join will cause the query to lose a LOT of performance. Not sure how you could measure this in mysql. Of course this was on large datasets.

ORs typically are very expensive.