Hi,

What I am trying to do:

User logs into a webapp running on server A. Server A contacts server B via HTTP, asks to download some persistent objects. Server B load()s those objects using hibernate, and returns them to server A (as Java serialized objects). Server A then persists those objects to its own database.

When the objects are first created/saved (on Server B), they get generated UUIDs. When they get saved to Server A, they need to keep their IDs so they have the same primary key on both servers.

There are plenty of object relationships which need to be preserved by the download.

My question is, are there any obvious reasons why this shouldn't work?

My first attempts have been unsuccessful -- i think i'm getting bitten by cascades. I'll try disabling them and see if that helps; still, any advice is appreciated.

Do you use the exact same mapping ?
Having the server A mapping with assigned id might help (beware saveOrUpdate and assigned).

HIbernate 2.1 offers the replicate() method for doing this stuff.

That looks like pretty much exactly what I want, but I'm having a problem with it.

I have a parent-child-type relationship mapped as a many-to-many using a List.

When I replicate() my top-level object, it puts rows for that object and all its children (recursively) into the database. So far, so good.

The problem is that it doesn't seem to be adding rows to the table which provides the many-to-many mapping.

It's clearly *finding* the reachable objects, since it's sticking them in the database. It just doesn't seem to be correctly replicating the association.

Is this intentional, or a bug, or do you think i'm doing something wrong?

-brian

Sounds like a bug. This is brand new stuff. Submit a simple testcase to JIRA.

I have opened issue HB-479 for this problem.

-brian

I have never used this way, but I think it must be more simple to send log file.
JDBC driver wrapper on sever A generates log file for updates with timestamps, server B connects, downloads, parses and executes sql.
Hibernate interceptor can help too, you can serialize objects in interceptor to log file, but I like text file more.

I have a problem with hibernate3.1 wihich looks quite similar.

I have a server running a oracle 10 database and many mobile clients using derby database, both using hibernate3.1 and the same database scheme (the same hibernate mapping file).

The mobile clients request objects (which are object trees) from the server database which were serialized and sent to the clients over network.

To persist the object tree on the client side I use Session.save on the root object since these object are new (transient) to the clients database.



All data is persisted to the db, except for the association tables of the many-to-many associations which are empty. On java side the association consists of a SortedSet implemented by TreeSet.

When I create a new TreeSet on the client using the TreeSet of the deserialized object everything works fiine for the many-to-many association between Catalog and Chapters:




In order to preserve primary keys all keys were generated by the application (generator="assigned").

Is this a hibernate bug or is it me doing something wrong here?