Comments on: How to select from an update target in MySQL

By: sql_developer

sql_developer — Thu, 14 Aug 2008 22:33:47 +0000

Thanks very much for this solution. I was having a hard time finding a portable way to do this.

By: Eyal Carmi

Eyal Carmi — Sat, 24 May 2008 13:18:52 +0000

Hi,
I want to raise a performance question regarding this solution and hope someone will have an answer to it.
I have a table with millions of rows, this table has a unique key: id.
For each row there are:
id, val_1,val_2, matchingRowId
I have already computed and updated the matchingRowId for each row and now i want to set each row’s ‘val_2′ based on ‘val_1′ of the mathcing row.

Originally i used:

update myTable set val_2=
(select T1.val_1
from myTable as T1
where T1.id=myTable.matchingRowId);

This of course failed but when i tried rewriting the query as you suggested to:

update myTable set val_2=
(select T1.val_1
from (select * from myTable) as T1
where T1.id=myTable.matchingRowId);

This works but the performance is terrible (since myTable is very large). I then read your second paper but i am not able to optimize the query since the following:

update myTable set val_2=
(select T1.val_1
from (select *
from myTable as T1
where T1.id=myTable.matchingRowId));

But got an error message saying that ‘myTable’ is not recognized in the inner loop.

Can anyone help?

Thanks

By: mebugus

mebugus — Wed, 21 May 2008 04:24:24 +0000

Only Thanks for this hint.

By: mysticav

mysticav — Mon, 21 Apr 2008 04:47:55 +0000

For a credit / debit/ balance table, this is an example:

INSERT INTO account set credit = 100, balance =ifnull((select balance from (select ((sum(credit)-sum(debit))) as balance from account) as x),0) 100

For instance, If you run twice the query, the table will be:

| credit | debit | balance
100 0 100
100 0 200

By: David

David — Wed, 09 Apr 2008 22:01:01 +0000

(^__^)
(_ _)
(^__^)
(_ _)

thank you x

By: Xaprb

Xaprb — Sat, 29 Mar 2008 15:03:20 +0000

On how to understand how MySQL works? The book I helped write, when it is published, will be the best. http://www.oreilly.com/catalog/9780596101718/ Note that that’s the second edition: the first edition is not as detailed.

Otherwise, I think Pro MySQL is a very good book, written by my friends Jay Pipes and Mike Kruckenberg. http://www.apress.com/book/view/159059505X

By: T1

T1 — Sat, 29 Mar 2008 13:54:31 +0000

What’s a good book on this topic?

By: Andy

Andy — Thu, 14 Feb 2008 20:28:10 +0000

This works great even for delete statements. I had a table with a linked id to another table but didn’t create a cascade foreign key yet and had some entries from the top table deleted. So to add a foreign key, I had to delete all the entries from the linked table that had nothing linking to it so I did this:

DELETE FROM tbldetails WHERE ID IN (SELECT ID FROM (SELECT tbldetails.ID FROM tbldetails LEFT JOIN tbltop ON tbltop.ID=tbldetails.topid WHERE top IS NULL) AS X);

(I replaced the actual names that would make more sense to what I’m talking about, hopefully). Although I would think this is a common thing and wonder if there’s an easier way to do that.

By: Daniel Barradas

Daniel Barradas — Wed, 28 Nov 2007 09:42:41 +0000

This worked for me.

I needed to update a number with the max value of a column in the same table.

update inscricoes
set dorsal = (
select dorsal from (
select max(dorsal) 1 as dorsal from inscricoes
) as dorsal)
where id = ‘$id’

Do you see any problem that may arise from this approach? (with concurrency maybe?)

Anyway … thanks for sharing! ;)

By: Xaprb

Xaprb — Mon, 20 Aug 2007 12:18:45 +0000

It looks like you can just rewrite that as a self-join instead of using subqueries. This will probably work much better:

update rehabs as t1
   inner join rehabs as t2 using(name, address1,...)
set t1.phone2 = t2.phone