Re: Delete cascade?
Re: Delete cascade?
- Subject: Re: Delete cascade?
- From: Chuck Hill <email@hidden>
- Date: Mon, 19 Mar 2007 16:48:35 -0700
On Mar 19, 2007, at 4:20 PM, Ian Joyner wrote:
On 20/03/2007, at 2:57 AM, Chuck Hill wrote:
On Mar 18, 2007, at 10:11 PM, Ian Joyner wrote:
What you say makes sense. Maybe, because my dependent entity does
not have any further cascades, I'm not seeing anything else being
brought into memory (maybe it's already there).
I think what I would expect is to see is:
SELECT * FROM <owned_table> WHERE OWNER_KEY = <PRIMARY_KEY of owner>
<processing of any cascades from those records>
DELETE FROM <owned_table> WHERE OWNER_KEY = <PRIMARY_KEY of owner>
That is not safe. What if records were added to <owned_table>
with an OWNER_KEY = <PRIMARY_KEY of owner> while you were
processing the cascades?
My unease is growing here, but then again I suspect it must be
right because it has been in operation for such a long time.
I suspect you might be misunderstanding the sequence of events.
I would guess that WO is handling the cascades due to underlying
DBs handling them in different ways.
No. EOF is handling them that way as it needs to be able to call
validateForDelete on the objects and may need information they
possess to process other delete rules.
I think records could be added to the <owned_table> by another
transaction from another user in any situation, unless maybe the
whole table is locked, or at least the parent record is locked in
such a way that owned records cannot be added. So you want the
DELETE to be atomic.
The delete _is_ atomic. Maybe hundreds of single DELETE statements,
but all within the context of a single, atomic DB transaction. The
DB transaction spans a shorter time time (read DB, process rules,
update DB) than the single, atomic EOF transaction. It is important
to keep in mind that the DB transaction is "nested" within the EOF one.
What you _don't_ want is to successfully delete the parent if someone
has just added children. Your suggested delete would do just that.
That might be acceptable in some situations, but not in all. EOF
pointedly deletes what it knows about and nothing else. If there are
other records, the DB will return an error when the transaction is
committed. This will inform the application which can then take
appropriate action.
I suspect that is why the designers of EOF elected to not use this
"optimization". Generally, they seem to have chosen safe over
fast in implementing EOF.
I agree with that, but I have a feeling that expanding the
transaction over time makes for more chance for untoward things to
happen. However, limiting the possibility for a race condition to
occur does not mean that it won't occur and have very nasty
consequences.
I don't think there is any race condition as EOF is implemented. One
process or the other will receive the correct error notification.
Thanks for your and Peter's answers, I don't want to waste anymore
of people's time on it, since the answer seems to be that it's
working as intended (and hopefully correctly) so it's nothing I'm
doing that's stupid to cause seemingly non-optimum (and I hope
safe) behaviour. I don't believe this is covered in the wikibook yet.
Learning more about how EOF works is rarely a waste of time. I doubt
that is in the Wiki.
Chuck
However, all I see is the DELETEs a single one at a time. Maybe I
just have an simple case?
Thanks
Ian
On 19/03/2007, at 2:08 PM, Peter Vandoros wrote:
Hi Ian,
WebObjects does this because it needs to bring all the EO's into
memory to process the delete rules specified in your EOModel.
Regards
Peter
Ian Joyner wrote:
I just noticed in my system that if I have the cascade delete
rule set on a to-many relationship in EOModeler that the
deletes happen one by one.
That is if I have a master record with three detail records, WO
issues three SQL DELETEs.
The setup is, each record has a primary key PRIMARY_KEY, which
relates to the owned records which have OWNER_KEY. To do the
delete cascade, I would think that DELETE FROM <owned table>
WHERE OWNER_KEY = <PRIMARY_KEY of owner> would suffice.
Have I got something set up wrong here, or am I missing something?
Thanks
Ian Joyner
Sportstec
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40etechgroup.com.au
This email sent to email@hidden
--This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
--
Peter Vandoros
Software Engineer
Etech Group Pty Ltd
Level 3/21 Victoria St
Melbourne VIC 3000
Australia
Ph: +61 3 9639 9677
Fax: +61 3 9639 9577
----------------------------------
IMPORTANT: This e-mail message and any attachments are
confidential and may be privileged. If received in error, please
reply to this message and destroy all copies and any
attachments. You should check this message and any attachments
for viruses or defects. Our liability is limited to resupplying
any affected message or attachments. For more information about
Etech Group, please visit us at http://www.etechgroup.com.au.
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40sportstec.com
This email sent to email@hidden
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40global-village.net
This email sent to email@hidden
--
Practical WebObjects - for developers who want to increase their
overall knowledge of WebObjects or who are trying to solve
specific problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
40global-village.net
This email sent to email@hidden
--
Practical WebObjects - for developers who want to increase their
overall knowledge of WebObjects or who are trying to solve specific
problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden