Rodrigo Senra wrote:
I guess what Chris meant was:

Given a pivot point in time (date and time), for *all* objects in the database (the whole Data.fs) will suffer UNDO over transactions whose timestamp is
 greater (more recent) than the given pivot timestamp.

Sounds about right...

Nevertheless, some doubts:

 - if you have several ZODB mount points, pointing
   to different files ? Should all be considered
   as a single database and suffer the same massive
   UNDO operation ?

Yes, although I don't have to worry about that for this project, since I only have one mount point to worry about ;-)

 - Should that mega-undo imply in a ZODB pack as well ?
   I suppose not.

Well, ideally I'm looking for non-transactional undo. ie: make the data "go away", not "write an undo record".

 - Wouldn't it be nice to figure out a way to specify
   a restricted scope for this operation ? For instance,
   give a relative root or set of meta_types ?

No, I want a quick, simple win here...



Simplistix - Content Management, Zope & Python Consulting
For more information about ZODB, see the ZODB Wiki:

ZODB-Dev mailing list  -

Reply via email to