This brings up an interesting point... WinCVS can *now* check out these Zope objects, but the toXML implementation makes them less useful than they possibly could be. For example: http://cvs.spvi.com/cgi-bin/viewcvs.cgi/zopeObjs/testCVS/hehe.xml.diff?logso... shows a 'diff' between two versions (1.10 and 1.11) of a DTML Document. It's nice since a small change in the document produced a 'small' change in the xml representation. However: http://cvs.spvi.com/cgi-bin/viewcvs.cgi/zopeObjs/testCVS/hehe.xml.diff?logso... shows a similar diff between 1.8 and 1.11, but now the diff is almost useless since the xml tags have all changed, confusing CVS. I snooped through the implementation, but I must confess, it was less than totally obvious if the 'ids' and such for the xml data were being used as convenient, clearly unique, but essentially random values, or if it was really important that they have exactly the values they did. Would it be possible to generate XML that was repeatable? (i.e, zope objects with exactly the same class and properties would always produce the same XML output?) Any Zope GURUs out there? ;-) thanks, -steve