Probably found the culprit.

I have added the appropriate attribute programmatically to the entity; and it 
seems that it was NOT added -- and therefore storedValueForKey did not find it. 
Makes sense. Oh, sigh.

All the best,
OC


On 12. 1. 2015, at 6:59, OC <[email protected]> wrote:

> Hello there,
> 
> I must be missing something very obvious now.
> 
> My application (rather complex) lately started sometimes to go stack 
> overflow. I have found the culprit is a code where I call storedValueForKey 
> in my own ERXGenericRecord subclass. The stack goes like this:
> 
> (i) handleQueryWithUnboundKey (which my EO overrides) get called -- I do some 
> internal checking there, and eventually call my own installed accessor, 
> essentially, 'this.foo()'. That works well.
> 
> (ii) the 'foo' accessor contains just "this.storedValueForKey('foo')", and it 
> indeed gets called all right.
> 
> (iii) now, I thought it would try to get the value of 'foo' from snapshot, or 
> fetch it from the database, or whatever. I was wrong -- it goes again to 
> handleQueryWithUnboundKey, GOTO (i), stack overflow.
> 
> Here's the important part of backtrack:
> 
> ===
> ...
>       at 
> cz.ocs.model.OCSEnterpriseObject.handleQueryWithUnboundKey(OCSEnterpriseObject.groovy:1173)
>       at 
> com.webobjects.foundation.NSKeyValueCoding$Utility.handleQueryWithUnboundKey(NSKeyValueCoding.java:494)
>       at 
> com.webobjects.foundation.NSKeyValueCoding$_KeyBinding.valueInObject(NSKeyValueCoding.java:894)
>       at 
> com.webobjects.eocontrol.EOCustomObject.storedValueForKey(EOCustomObject.java:1634)
>       at 
> com.webobjects.eocontrol.EOKeyValueCoding$storedValueForKey.call(Unknown 
> Source)
>       at cz.ocs.model.OCSEOUtilities$FOO_ACCESSOR(OCSEOUtilities.groovy:60)
> ...
>       at 
> cz.ocs.model.OCSEnterpriseObject.handleQueryWithUnboundKey(OCSEnterpriseObject.groovy:1173)
> ...
> ===
> 
> It even is documented thus, EOCustomObject.storedValueForKey tries first _foo 
> and foo, then... yadda yadda ... and if all fails, it is documented to go 
> handleTakeValueForUnboundKey (which seems to be a typo in documentation, 
> meaning handleQueryWithUnboundKey instead):
> 
> ===
> ...
> public Object storedValueForKey(String key)
> Returns the value for the property identified by key. This method is used 
> when the value is retrieved for storage in an object store (generally, this 
> is ultimately in a database) or for inclusion in a snapshot. The default 
> implementation provided by EOCustomObject is similar to the implementation of 
> valueForKey, but it resolveskey with a different method instance variable 
> search order:
>               • Searches for a private accessor method based on key (a method 
> preceded by an underbar). For example, with a key of "lastName", 
> storedValueForKey looks for a method named _getLastName or _lastName.
>               • If a private accessor isn't found, searches for an instance 
> variable based on key and returns its value directly. For example, with a key 
> of "lastName",storedValueForKey looks for an instance variable named 
> _lastName or lastName.
>               • If neither a private accessor or an instance variable is 
> found, storedValueForKey searches for a public accessor method based on key. 
> For the key "lastName", this would be getLastName or lastName.
>               • If key is unknown, storedValueForKey calls 
> handleTakeValueForUnboundKey.
> ...
> ===
> 
> But darn, _where_ then is the specific EOF code to get the attribute value 
> from snapshot/from database? So far I thought EOCustomObject overrides the 
> generic storedValueForKey appropriately (so that instead of trying _foo, 
> _getFoo, foo and getFoo it reads from snapshot/database). It does not seem 
> to, though?!?
> 
> What am I missing here?!?
> 
> Thanks a lot,
> OC
> 


 _______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list      ([email protected])
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/webobjects-dev/archive%40mail-archive.com

This email sent to [email protected]

Reply via email to