Hi Ivan et al, I would like to make sure that I understand your intention.
(a) JCR should not mandate the scope of query wrt the transient space (b) Jackrabbit should implement isolation differently In case you are interested in option (a) then I would like to suggest to send this to [EMAIL PROTECTED] I guess that from the discussion on this list it is relatively evident though that there are a number of arguments that can be made for either way implementing or using a content repository. I think you have a point that if one can argue both ways it probably makes sense to allow both implementations in the specification, and therefore rewrite the section you quoted in the beginning of this thread. Again, please feel free to suggest that to [EMAIL PROTECTED] If you comments are more targeted at (b) I think [EMAIL PROTECTED] is the correct audience, and a patch would certainly be a good start ;). Personally, I believe that the isolation levels problem particularly when it comes to things like ranking in full-text search or transient modifications that do not satisfy nodetype constraints can become more tricky than expected. regards, david
