On Dec 14, 2010, at 1:58 PM, René Bock wrote:
Hi all,
After migrating our CMS from WO 5.3 to 5.4 we noticed some strange
effects on the PrimaryKey generation via EO_PK_TABLE
Our project makes extensive use of horizontal inheritance, using
different models in different frameworks.
Normally, there is (and should be) only on row per base entity in
EO_PK_TABLE. But this rule was broken, when we try to call
anEo.permanentGlobalID() for an unsaved eo.
It has the effect, that now in EO_PK_TABLE there is an entry for
anEo.entityName(), but the pk value of the base entity was not
incremented (which leads to not so amusing effects)
As, of cause, I can't reproduce this behavior with a simple test
application, I would like to here if anybody else has experienced
this phenomena or hears ringing a bell?
I know OpenBase fails with HI for this same reason. In the openbase
plugin, the newPrimaryKeys() is incorrectly implemented and uses
entity.externalName() to find the table name for key generation. It
should use entity.primaryKeyRootName() instead. As a result, the leaf
tables provide the primary key producing collisions.
Luckily I can avoid calling permanentGlobalID on unsaved EOs, but
not knowing why pk-generation may be broken, doesn't let me sleep to
well...
I'm using WO.5.4.2, current Wonder _5_4_branche and using either
MySQL or Microsoft SqlServer...
I agree with Pascal. You should definitely update to WO 5.4.3.
Ramsey
regards,
René Bock
Software Engineering
--
Salient GmbH - the interactive advantage
http://www.salient.de http://www.openforms.de
_______________________________________________
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]
