WhenSight is a 3D Visiualization tool for time-aligned information. Screen Shots:

WhenSight / StackTrack integration - "Starved Phil" Example from Boost::Threads

WhenSight PDH (Performance Counters)

---

Searching for time-based relationships and behavioural patterns in large data sets can be challenging. Some of the major issues:

Clear view of the "forest"

Large volumes of data must be presented in a single view without overwhelming detail

Disparate data

Multiple kinds of events kinds with little in common other than time must be presented in a cohesive manner

Real-Time display

It is often important to monitor "live" activity. While there is value in creating static representations of pre-existing data, the ability to have time-oriented data tracked as it happens can be a significant advantage.

Varying Intervals

Representing order (a, then b, then c, then d) is insufficient. It is important to convey, for instance, that there was ten seconds beween a and b, 0.5 seconds between b and c, and 10 seconds between c and d. Order-based representations can be useful with a consistent time series, but are poorly suited to data with irregular intervals.

The current implementation of WhenSight is based on the following components:

C++ (MS VC7.1)
Python 2.5
OpenSceneGraph 2.2
Python OSG bindings based on Py++
Boost C++ Libraries (1.34.1)