Major resource leak

[kraney]

All of the fusion charts we're using have a problem where, after being used for a while, the chart will "stick" showing old data, and the browser will start to consume 100% CPU resources.

 

 

 

I have a license including source code. I am able to identify the source of this problem.

 

 

 

The problem is due to a recurring usage pattern with the setInterval call. Most if not all of the charts make calls to setInterval during rendering, saving the interval to a class member variable. Later, the method being called will call clearInterval passing the class member variable.

 

 

 

This is not reentrant.

 

 

 

The problem occurs when a call to reset the chart occurs *before* the interval fires. This causes setInterval to be called again, and the interval to be saved to the class member. The previous interval is still active, but is lost. When the method is called by the interval, it will clear the second interval, but the first one will be left to trigger over and over.

 

 

 

I would expect similar issues if the chart is destroyed before the interval fires.

 

 

 

After one occurrence, this will cause rendering issues since the old call will keep occurring. After several occurrences, the leaking intervals will pile up, consuming CPU resources.

 

 

 

 

 

To fix the bug, it would be necessary to prevent a new interval from being set if one is already in place.