Thanks for your help Chuck,
Been there...
Glad to hear I'm not dreaming this.
I have seen this in these cases: 1. Use of unlocked editing context
see below
2. Use of mutable attributes
I'm 100% certain this is not the case.
3. Faults firing during saveChanges (while processing deletes in particular)
Everything happens in one EC. The order shouldn't be a fault at this point, as the order was created and saved in it already. But nevertheless, my stomack says, this is the route to follow. I keep the session alife indefinitely using ERXJSLifebeat, and I'm under the impression the problem happens more often after either app-startup or a long period of idle time. Maybe something getting garbage collected/ invalidated after a while?
Is there any chance that autolocking is turned off for ERXEC?
I'm pretty sure it's not, but I'll add more checks. The annoying part of the problem is, that it's so hard to reproduce. I've setup Selenium tests, creating 100 orders each, but none of the tests exhibits the problem. In production I get around 10 "bad orders" a day, while 50-300 orders work fine. This morning I could immedeately reproduce the problem 2 times in a row and already felt on the winning road, but after that, it took hors to get the next failure...
Chuck
Thanks for your help. I owe you some gallons of beer at WOWODC. :) Timo _______________________________________________ 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]
