[JBossCache] - Re: TreeCache/Hibernate/JBossAS optimistic locking problem
by steve.ebersole@jboss.com
I have some local mosds to these tests...
I just confirmed with Manik regarding the mixing of calls supplying and calls not supplying a data-version. Hibernate will need to provided data-version info for all nodes under its control reagrdless of whether optimistically locking these nodes makes any sense...
Reagrding the warning, the message is just a bit misleading (and I have actually changed it locally). It is really protecting that *inserting* data is not involved in a lock check (because that would mean that data already existed for the entity we are trying to insert). But I found an interesting thing (which perhaps was the initial thing which caused me to add this): basically, after an insert in one transaction, the node is validated against itself in the next transaction!?!?! I would guess that it has something to do with the fact that tree cache is registering its synch inthe midst of a beforeCompletion() cycle, but not certain. Would anything else cause tree cache to validate a non modified node in this manner?
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3957442#3957442
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3957442
19 years, 9 months
[JBoss Seam] - Re: Beginning conversations in @Factory methods
by iradix
anonymous wrote : "petemuir" wrote : anonymous wrote : Is there no way to end a current conversation and begin an entirely new one in the course of a single request?
| |
| | I'm pretty sure there isn't, Gavin wrote about this sometime ago on the forum.
|
| I searched the forum and hadn't run across anything. I'd be interested to know why this is the case. Even if it's not the default behaviour a @End(immediate = "true") feature seems like it would be useful. It would also help if sometime in the future the conversation lifecycle was documented in a bit more detail, although I realize it's a work in progress and that might take a while.
|
| anonymous wrote : AFAIK you can use the back button when using jpdl:
|
| Thanks for the tip. I hadn't noticed when that feature was added. Any idea how it works? Some type of unique token representing the current state that's placed in the component tree perhaps?
|
| anonymous wrote : What about putting the datamodel into the PAGE scope? Then start a conversation on the next action.
|
| My problem is that the next action after a list is displayed is often displaying a selection from the list. As I understand things, on a postback the DataTable component will reaccess the DataModel through EL variable resolution, then indicate the row selected. If the DataModel exists in a long running conversation, that will be returned and it'll work great. If the conversation wasn't started when the original DataModel was generated I'd imagine a new DataModel will be created on the postback (through the @Factory method) and it could be different from what was rendered to the user. Does that make sense?
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3957441#3957441
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3957441
19 years, 9 months
[Remoting] - Re: securing servlet invoker?
by tom.elrod@jboss.com
The problem with the servlet invoker on the server side is that because it is deployed in a web container, remoting does not know all the configuration associated (such as ssl). Therefore, just assume lowest common denominator, which is plain old http (non ssl). This is really only important for discovery since locator url published in detection message would be something like (servlet://myhost:8080/bla).
However, if are not using remoting discovery and can explicitly set the locator url on the client, then can use 'http' or 'https' as on the client side will be using the HTTPClientInvoker (or HTTPSClientInvoker) regardless of if is talking to http server invoker or servlet server invoker (which would be running within a web container). The HTTP(S)ClientInvoker has no implicit knowledge of what implementation it is sending and receiving http requests to/responses from.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3957434#3957434
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3957434
19 years, 9 months
[JBoss Seam] - Re: Beginning conversations in @Factory methods
by petemuir
anonymous wrote : Is there no way to end a current conversation and begin an entirely new one in the course of a single request?
I'm pretty sure there isn't, Gavin wrote about this sometime ago on the forum.
anonymous wrote : I'm not using jPDL pageflows (I need to support back button usage)
AFAIK you can use the back button when using jpdl:
| <page name="..." view-id="..."back="enabled">
| </page>
|
anonymous wrote :
| It makes sense to me that a conversation should be started when a @DataModel is accessed so that the corresponding @DataModelSelection matches up to the DataTable presented to the user. If the conversation didn't begin until a row was selected I'd imagine the postback would start a new conversation, retrieving a new List via the factory method with potentially different results. Or am I missing something about the conversation lifecycle again?
|
What about putting the datamodel into the PAGE scope? Then start a conversation on the next action.
View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=3957431#3957431
Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=3957431
19 years, 9 months