[IGSTK-Developers] Some comments after todays Tcon

[Frank Lindseth]

Thank you all for the Tcon discussion today.

Regarding the Tracker/TrackerTool setup, I hope you have the chance  
to spend a few minutes on the pseudocode below:
http://public.kitware.com/IGSTKWIKI/index.php/ 
TrackerToolExamples#Pseudocode_for_setting_up_a_tracker_and_some_tracker 
_tools
not many sentences there, but believe me they are well funded...

Regarding the use of the re-factored components (SpatialObject,  
Tracker, etc.), please have a look on the following pseudocode:
http://public.kitware.com/IGSTKWIKI/index.php/ 
TrackerToolExamples#Discussion
The NEW ImageResliceRepresentation was mentioned today, this is an  
example of how the "complete" picture COULD look like,
does this look ok, something that we could work towards, or is it way  
out of line, please comment.

I just did a "Recent changes" update on the IGSTK-wiki and found  
these additions:
http://public.kitware.com/IGSTKWIKI/index.php/Minutes_102507
http://public.kitware.com/IGSTKWIKI/index.php/ 
TrackerToolExamples#Polaris_Tracker_Tool_State_machine_diagram
http://public.kitware.com/IGSTKWIKI/index.php/ 
TrackerToolExamples#Aurora__Tracker_Tool_State_machine_diagram
I'm not sure that I completely understand the last two diagrams but  
based on the figures I realize that we do think a bit different about  
this.
I have many comments to this, but I just like to say that if we start  
out with a state machine that is this complex JUST for the setup  
(have we stopped adding all the "AttemptingStates" to the diagrams?)  
we will have a hard time completing it.

I'm not convinced that a "separate" state machine is needed for each  
Tracker/TrackerTool subclass, the code for getting from one "general"  
state to another will of course be different, at least for the  
NDITrackers and the MicronTracker. Given the fact that all NDI  
trackers share the same API it's not unnatural to think that a  
NDITracker(Tool) (subclass of Tracker(Tool) would capture most of the  
variability (maybe not all, so we might need AuroraNDITracker(Tool) /  
PolarisNDITracker(Tool) also). I don't think that a sharp distinction  
between wired and wireless tool are needed.

Ziv suggests specific Tracker(Tool) setup and general/abstract use  
(which I totally agrees with):
http://public.kitware.com/IGSTKWIKI/index.php/TrackerToolExamples#Ziv
will this be possible with the new implementation?

For the CoordinateSystem changes we have a specific "design" document  
that implementation could be based on:
http://public.kitware.com/IGSTKWIKI/index.php/Image:IGSTK- 
CoordinateSystem-C.pdf
Do we have something similar for the Tracker/TrackerTool re-factoring.

Regards,
Frank