Zope 2.1.0 beta 1 caution - please read
Hi all - We (DC) have run across a bug in the 2.1.0 beta that has given us some grief on the zope.org site, and I wanted to post a caution to those folks currently using the beta. In short, if you are doing any ZClass development where your ZClasses are based on Python base classes that are still in development (e.g. you are updating the Python base classes at times), there is a risk of getting a non- unpicklable object stuck into your object database. While the sequence of events required to cause this is pretty arcane, it can potentially be a real pain to recover from. So, if you are doing active ZClass development in 2.1.0b1 we strongly suggest that you either go back to 2.0.1 or take a break until tomorrow or Monday (I hope to have a 2.1.0 b2 ready that resolves this by then). If this is too painful (or you're hardheaded or just like living on the edge :) at least be sure to back up your Data.fs regularly just in case. Brian Lloyd brian@digicool.com Software Engineer 540.371.6909 Digital Creations http://www.digicool.com
On Thu, 11 Nov 1999, Brian Lloyd wrote:
Hi all -
We (DC) have run across a bug in the 2.1.0 beta that has given us some grief on the zope.org site, and I wanted to post a caution to those folks currently using the beta.
Oh dear - so that's what I did! To provide further info on this bug - it occurs when a Python base class defines an attribute the same name as an attribute / property of the ZClass based on it. What *appears* to happen is that it attempts to pickle the base Class attribute with the result that the database is corrupted. The following methods of recovering data don't work: * UNDO the transation - either I packed it or it packed it for me * DELETE the instances - Zope segfaults * ACCESS the instances - Zope segfaults * RECOVER (by shifting dates or whatever) the data - doesn't work So, how do I fix my database?? Cheers, Anthony Pfrunder
participants (2)
-
Anthony Pfrunder -
Brian Lloyd