are the workitem and the expression storage sharing the same JCR
workspace ? (Sorry never had the time to actually look at how the
separation is done in Magnolia).
Let's be sure I am not mistaking:
- config workspace: the flow definitions
- Expressions: the expressions
- Store: the workitems
As a side note, the flow definition path is hardcoded, so it's hard
to define new ones. Philipp was suggesting to add the "commands" node
for each module, I now suggest to add a new flow one. And to have
that ObservedManager running for flows.
A bit of context : when the OpenWFE engine evaluates a participant
expression, it stores the 'state' of the participant expression just
before delivering the workitem.
Ok.
In the MgnlParticipant class (that extends the
AbstractEmbeddedParticipant) the consume method does:
First, get the participant name.
Them,
-> in case of a command:
* get the command
* create a workitem context from the current context (but I
think that part of the code is wrong. It calls MgnlContext.getInstance
() but nothing is set since we are in a flow)
* execute the command
* reply to the engine
-> in other cases
* store the item (no reply?)
Does that sound correct to anyone ?
It seems that the further storage of the workitem (JCRWorkItemStore)
is complaining about the fact that the JCRExpressionStore hasn't yet
stored the 'participant' expression.
Ok, this is beyong my friday evening brain I guess ... :P
But I am listening to any comment and can work on it.
Niko,
----------------------------------------------------------------
for list details see
http://www.magnolia.info/en/magnolia/developer.html
----------------------------------------------------------------
