[prev in list] [next in list] [prev in thread] [next in thread] 

List:       ojb-user
Subject:    ODMG transaction abort bug?
From:       Saleem Shafi <saleemshafi () gmail ! com>
Date:       2004-10-29 20:45:54
Message-ID: f7b4029104102913454e73b57 () mail ! gmail ! com
[Download RAW message or body]

Hello,

There seems to have been some discussion on this issue, but i haven't
found a resolution yet.   I'm working off of OJB 1.0.1, and i'm
getting some undesired behavior when aborting a transaction.  I've got
a situation where PersistentObject A has a reference to
PersistentObject B.  I start a transaction, read A, navigate to B,
change B, then throw an exception.  The transaction aborts, walks
through the objects in the transaction, but only drops B from the
cache.  A is still around, so when i start a new transaction, read A
and navigate to B again, all the changes i made in the previous
transaction are still there.

I understand that actually rolling back the state of the objects is a
pain, but i'm thinking the issue can be pretty easily avoided by
simply removing all of the objects that were included in the original
transaction from the cache.  That way when i re-read A, i'll get a
fresh copy and also a fresh copy of B.  Obviously that means fewer
cache hits, but i vote that keeping the integrity of the transaction
sound is more important.


Saleem.

 
ps. i believe Matthias Laube brought the issue up first:
http://nagoya.apache.org/eyebrowse/ReadMsg?listName=ojb-user@db.apache.org&msgId=1641493

-- 
Saleem Shafi
Partner
Pomatics Technology Partners LTD
http://pomatics.com
ph: 512.577.1472
email: saleemshafi@pomatics.com

---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-user-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-user-help@db.apache.org

[prev in list] [next in list] [prev in thread] [next in thread]