(Steve, sorry, you will get this mail twice... it seems that the gmail reply didn't do what I was expecting ie posting to the mailing list)
--------------------------------------------------- Thanks Max for your comment on the forum, I was eagerly reading the thread in the mailing list and will now take time to add some water to the mill. Gavin, do you think it could be possible to downmerge in H2 you improvement on read-only objects since it seems something was done in H3? (worst case, if you give me some pointers I might be able to provide a patch for H2) About the second part, I understand it is not as easy to do as it easy to say. I will suggest to do an easy benchmark. Lets say you put around 10, 100 and 1000 objects in cache. You modified some ot them, sometime objects that are queried, sometime not. Then you perform a bunch of queries on them. You can do a quick hack just to handle basic mapping (nt cascade, collections or other tricky things) with the algorithm turned update down. Just to see if i worths the refactoring effort. But it's true that checking dirtiness on objects in the query table space and then complete then flush if needed is not that obvious. Maybe clearing the partial flush done on query table space and doing a flushEverything if it was determined that a flush is needed will still be faster and be easier. About the post-scriptum part on the forum. Using the code enrichment at compilation that H3 is doing can allow also to monitor dirtyness. That might also be a big performance improvement since no attributes compare will be needed anymore during the flush. Thanks for considering my comment, Henri P.S.: When doing benchmarks on hibernate, do you have a way to bypass the database so you can bench only hibernate performances? P.P.S: On thing I forgot. We are already doing http://blog.hibernate.org/cgi-bin/blosxom.cgi/2004/06/23#i18n. But by reference data I was meaning more complex objects (trading accounts, financial instruments, clients, stock exchanges). Things that might change lets say once a month. So we are willing to have them read-only most of the time and to refresh the level 2 caching manually when they are updated. On Tue, 4 Jan 2005 09:01:43 -0600, Steve Ebersole <[EMAIL PROTECTED]> wrote: > Right. Apparently I just used poorly chosen wording. What I meant to > say is that it is "conceptually" what TopLink achieves with its > separation of Session and UnitOfWork. As Gavin pointed out in his > response, sharing actual instances between sessions and/or threads is > not desirable. That was not my intent to imply such a thing. > > -----Original Message----- > From: Joshua Davis [mailto:[EMAIL PROTECTED] > Sent: Monday, January 03, 2005 10:18 PM > To: Steve Ebersole > Cc: Max Andersen; Hibernate development > Subject: Re: [Hibernate] someone looking at flushing performance... > > Steve Ebersole wrote: > > >[...snip...] Then we could continue to track these > >objects within the scope of object identity but not in the session > cache > >(aka, this is analogous to TopLink's distinction between Session and > >UnitOfWork). > > > > > ... except better than TopLink. The TOPLink Session stores objects > that are used by multiple threads, which introduces a lot of problems > that nobody needs to have. > > > > >-----Original Message----- > >From: [EMAIL PROTECTED] > >[mailto:[EMAIL PROTECTED] On Behalf Of Max > >Rydahl Andersen > >Sent: Sunday, December 26, 2004 1:53 PM > >To: Hibernate development > >Subject: [Hibernate] someone looking at flushing performance... > > > >...fyi: http://forum.hibernate.org/viewtopic.php?t=936291 > > > >Is he on to something or is the readonly stuff in H3 the answer to this > >? > > > > > > > > ------------------------------------------------------- > The SF.Net email is sponsored by: Beat the post-holiday blues > Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. > It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt > _______________________________________________ > hibernate-devel mailing list > hibernate-devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/hibernate-devel > ------------------------------------------------------- The SF.Net email is sponsored by: Beat the post-holiday blues Get a FREE limited edition SourceForge.net t-shirt from ThinkGeek. It's fun and FREE -- well, almost....http://www.thinkgeek.com/sfshirt _______________________________________________ hibernate-devel mailing list hibernate-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/hibernate-devel
