On 01/04/2009, at 8:58 AM, Mike Schrag wrote:
Careful, that VI stuff can be dodgy.
As it stands this is true with the stock frameworks - though the
fixes are easy for Apple to implement. Now, alas, there's a
regression in 5.4 with VI too .. but I think it'll only get fixed
in EOAccess when I set up a test case with Apple's own inheritance
example app and send it back to them. Gotta love the irony here ;-)
definitely log a radar
I've certainly logged radars about any of this stuff.
[1] 6592526 (17 Feb 09) EOF Vertical Inheritance now inserting
duplicate pks: 5.4 regression with pk generation for VI as mentioned.
[2] 6302622 (20 Oct 08) NeededByEOF<index> not resolving for
EOEntity.anyAttributeNamed - awaiting test case from me admittedly
[3] 5632577 (06 Dec 07) lock for update on pk operations
[4] 5219252 (22 May 07) EOF fails to generate foreign key in sql
(vertical inheritance)
Notes:
[1] fix is now in wonder
[2] fix is in wonder
[3] Worked around in my EODatabaseContext subclass. The problem is
it's creating a lock for update on parent pk rather than lock for read
while child entity is being updated. There doesn't appear to be a
'lock for read' database operation in eof. So the real fix doesn't
appear to be simple even though my own work around is one line of
adjustment code... but there's too much private copying from
EODatabaseContext to share this. There's no hook for this to override
in other words without hackery.
[4] This is where unless in the EOModel the foreign key of a child
entity needs to be exposed in Entity Modeler as otherwise generation
fails at runtime. Work around is to expose the fk as a class attribute
in Entity Modeler
So apart from '3' I've submitted everything - and you've got 3
privately if you've got any thoughts :)
with regards,
--
Lachlan Deck
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list ([email protected])
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com
This email sent to [email protected]
