Christoph Reck wrote:
>
> If I get this right, VM are currently allways global?
Yes, they are currently accessable by all templates.
> This whould be another reason for making VM access into
> references (using '$' and not '#'). They should be placed
> into something called a 'base context' which is cloned on
> each request. Local VMs are then volatile!
Noooooooo. :)
> I believe some documentation is needed on the context of
> a VM:
> * Where does a VM live - in the context or in the
> directive look-up-table?
Neither. They have their own home, which you can consider to be like
their own directive look-up table...
> * What references/variables does it see internally -
> only the passed ones or all of the current context?
They have access to the current context like any other piece of VTL.
> * What happens with local #set's, are they visible
> after the VM returns?
Of course. Like any other VTL. (It's not really a subroutine)
> * What happnes with local #set's when they have the
> same name as in the global context?
There is no 'global context' per se. Generally, each is associated with
the thread of execution using Velocity. (Right : most people will do a
Context c= new Context() in their own code, or the servlet base class
does it for them...) Right now, it's expected that the Context object
is per-thread. In the current implementation, there might be problems
if you share the Context across threads. (Same with WM).
> * It seems that it is strongly recommeded to place
> VMs into libraries, which are loaded on startup.
Yep!
> Local VMs may be used within a template if the
> configuration allows it.
Yep!
> The issue of this thread
> is to clarify if these VMs live past this template
> access and are accessible from other templates.
Yep! They currently do. I pretty much think it would be straight
forward for me to put in a switch (property) that will allow VMs
specified in a template to be only visible to and used by that template.
(And we ain't making them references...:)
geir
> Thanks in advance for clarifications.
>
> :) Christoph
--
Geir Magnusson Jr. [EMAIL PROTECTED]
