[IGSTK-Developers] Some discussions for today's Tcon

[David Gobbi]

Hi Patrick,

I'll add these four issues to our weekly agenda so that the group will 
keep on reviewing them until they get fixed.  If you could add each of 
them as a bug report, that would also help.

See additional comments below:

Patrick Cheng wrote:
> Hi Guys,
>
> Some major issues I had for the demo. I put some fix in the 
> Sandbox/IGSTK/Source directory, but I want to talk about it today and 
> ask for your advice.
>
> 1. TimeStamp unite inconsistency problem
>    a. RealTimeClock get methods return value in seconds, while we are 
> using milliseconds to setup expiration time.
This is definitely something that needs to be changed.  I looked at the 
code and RealTimeClock::TimeStampType is seconds while 
TimeStamp::TimePeriodType is in milliseconds, but both types are 
"double" and if programmers mix and match them the compiler won't flag 
an error.
>    b. confusion in coding 1, unit
>       eg. in igstkView.cxx
>       renderTime.SetStartTimeNowAndExpireAfter( 1 / frequency );
>       should be:
>       renderTime.SetStartTimeNowAndExpireAfter( 1000.0 / frequency );
>    c. coding confusion 2, duration, not time
>      eg. in igstkTracker.cxx
>       toolTransform.SetTranslationAndRotation(translation, rotation,
>                           transform.GetError(),
>                           transform.GetExpirationTime());
>       Should be:
>       toolTransform.SetTranslationAndRotation(translation, rotation,
>                           transform.GetError(),
>                           (transform.GetExpirationTime()-
>                           transform.GetStartTime())*1000);
To enforce the difference between "absolute" time and "duration" time, 
we might need to introduce a new class type for each.  These classes 
would just be a container for "double", but they would enforce correct 
usage.
> 2. Constant & Dynamic transform implementation problem, and transform 
> multiplication
This has been discussed, we need to revisit it.  IGSTK needs some way to 
recognize certain transforms as non-expiring.
> 3. ObjectRepresentation state machine refactoring. TimeStamp is not a 
> property of represenatation class, we should eliminate the following 
> states.
>   //igstkAddStateMacro( ValidTimeStamp );
>   //igstkAddStateMacro( InvalidTimeStamp );
>   //igstkAddStateMacro( AttemptingUpdatePositionAndVisible );
>   //igstkAddStateMacro( AttemptingUpdatePositionAndInvisible );
>
> 4. TubeObject problem, Is the geometry really in SpatialObject class? 
> We have a problem here.
I'm not familiar with these two issues, so I will leave them for other 
people to discuss. 

 - David