[Paraview-developers] Updating view properties

[David Thompson]

Hi Utkarsh (et al.),

I had to work on some other stuff for a while, but I've managed to  
create a small example of the problem. Attached is a tarball  
containing a ParaView plugin that implements a new view type: "Dummy".  
It works when you compile as is, but changing the #defines in Dummy.h  
will create a version where the server exits on any panning/zooming of  
the plot.

	Thanks,
	David

On Feb 11, 2011, at 6:00 , Utkarsh Ayachit wrote:

> If you have some code that I can play with, it will be easier :).
>
> Utkarsh
>
> On Thu, Feb 10, 2011 at 8:48 PM, David Thompson <dcthomp at sandia.gov>  
> wrote:
>>>> No. Interaction (panning/zooming) on the client does not result in
>>>> Render()
>>>> being called on either the client or the servers.
>>>
>>> Argh..that's what I feared. Okay next step. So you have a
>>> vtkPVContextView subclass right?
>>
>> Yup.
>>
>>> In that subclass, add an observer to the vtkContextView's EndEvent  
>>> (or
>>> whoever is firing that event). In the callback, call:
>>>  vtkMultiProcessStream stream;
>>>  stream << this->Identifier << static_cast<int>(blockindex);
>>>  this->SynchronizedWindows->TriggerRMI(stream, SOME_TAG);
>>> This should result in the server-side processes reporting that an
>>> un-handled RMI was invoked.
>>
>> Not quite. When I add this (switching out blockindex for a couple  
>> doubles of
>> interest and SOME_TAG for my favorite random integer constant), it  
>> gets an
>> error which would result in a "Could not receive RMI trigger message"
>> message within vtkMultiProcessController::ProcessRMIs except that
>> reportErrors is hardwired to 0 by the caller.
>>
>> Interestingly, if I force the client not to call TriggerRMI for the  
>> first
>> few times (until the first rendering is done), then I do get the  
>> message
>> about the unhandled RMI each time the client renders the chart. So  
>> it seems
>> like the first calls on the client are occurring in a bad spot.
>>
>>> If this works correctly, then you can add code to handle this RMI  
>>> on the
>>> server-side and in that you can change the ivar on your  
>>> vtkPVContextView
>>> subclass.
>>
>> I think I see how to do this... I just want to get the point where  
>> the
>> server doesn't jump out of the car while we're turning a corner.
>>
>>        Thanks,
>>        David
>>
>>
>