################################################ # To Do ################################ refactor to use MotionModel.addTemporalVariable curve brake graphic (if easy) fake 3d display for showing platform mass >>Current implementation has problems; not sure this will work • In-Chart readout for acceleration go offscreen when showing all 3 components at 1024x768 • Start paused, use wiggle me for startup (low priority, don't implement yet) • add help button if the sim ever has help • Play area layout is wrong under linux • Chart maximize buttons disappear under mac Performance Issues • Improve performance? • The “speed of time” going by changes significantly if the disk is spinning or not. • The speed of the sim shows surges pretty regularly … not smooth motion even on my fast processor. >>These issues should be resolved due to clipping the play area and using the improved renderer for graphics. • There is still a performance leak. After about 10-15 minutes it is now crawling. >>Is this fixed now? • Crashed after an hour or so, out of memory error • Be able to switch between radians and degrees? >>Eventually we'll want to redesign the charting library so that this feature is built-in, but this will take some time. • I think we need to think about the graph options we have (and what we let them control). … Should we actually revisit relation between angle, angular vel, and angular acc. … or just relationship between angular acc, torque, and moment of inertia. But tricky part is that torque is due to applied force so need to control force and radius in a controlled way to explore this relationship. Should this torque sim get into relationship between angular momentum and angular velocity, or should this be another companion sim? ################################################ # Questions/Discussion ################################ • the theta = 0 indicator is partially obscured by the platform handle on rotation panel. What shall be done to fix this? • Some vectors go out of the play area, should we shrink the vectors or zoom out more? • I like that the friction force looks sandwiched between the block and the platform; however if the force is too small, this is difficult to see. Should we move this to the front? • What colors should be used for forces/torques/radii in the torque simulation? • Add a visualization for the platform mass? Perhaps the platform should have a thickness displayed? • Bug should fly off at high speed (if not too much work) >>Probably 1.5-2.0 hours, should we go for it? • Support for "missing values" in the series data (for when characters are being dragged by the user and don't have values for angular data series) -7-19-2007: No missing values now -Current implementation linearly interpolates between previous and subsequent values -It would be fairly difficult to add a "missing value" feature -We could also consider just computing these values dynamically • There is still a very large delay on the v and a vectors when dragging the bugs around on my machine it makes it look like the arrows are pointing outwards for acceleration, not inwards when I move in a circle by hand. -Current implementation uses a mix of circular regression and 1-d derivatives with time delay; let me know if this is sufficient. >>Changed thresholds so motion should be classified as circular more accurately • When you set angular position by hand … currently only lady-bug position affected… should we make similar change in theta to 2nd bug if he is on graph? So if ladybug is at 100 rads and other bug is at 110 rads, then set rads to 0, end up with ladybug at 0 rads and other bug at 10 rads??? NOT sure about this? >>No, I think the settable values should only apply to the ladybug. It will be too confusing/overloaded to make it so you can type in values >>for both characters. • When ever I move the bug around in what seems to me to be a circle the acceleration points outward. >>Did performance improvements fix this problem? >>I can relax the conditions that must be met to recognize circular motion. • The delay in the vectors between when I let go and when the vectors stop moving. >>Did performance improvements fix this problem? The beetle graphs show up when you put him on the platform. He can have been off the entire time and then when I put him on, his data appears. When I take him off, his data disappears. This is odd. I can see why new data won’t graph when he’s off but not why all his data comes on and off depending on when he’s on the platform. This gives the graphs the feeling that they are a physical thing rather than an abstract representation of his motion. >>Perhaps as soon as the user grabs the second character, his graphs should become visible (and his checkbox becomes selected). >>Then to hide the graph, the user unchecks the checkbox. >>The idea is that as soon as the user expresses an interest in the beetle, its graphs become visible >>As opposed to our previous version, which assumed the user only wanted to see data if the bug was on the platform • I want to discuss the beetle graph behavior with Kathy before we do anything new with this. I can see there are some tradeoffs and lots of scenarios that need to be considered. • I think the default axis for the velocity (linear) graph could be different but again here I think I should discuss with Kathy first. ################################################ # Lower Priority ################################ • Bug: Mouse gets disassociated from ladybug character after trying to drag out of bounds. • Improve correspondence of graph time scale with wall-time clock speed • Shadow projections should be considered after AAPT. • Possible interview question: bug icon at tip of graph to improve conceptual connection • Torque mode: Should be able to "spin" the wheel (i.e. drag, let go, and have it continue to spin, at least in friction mode)? • some pswing components are fuzzy -can't reproduce this on development machine • typing in a box without pressing enter should work -can't solve this in general, since we don't know when the user has finished entering input • Bug: Beetle graph appears empty before beetle is on the platform, why isn't this interpolating like the other graphs? -values for RotationBody's angleSeries aren't set while off platform • Consolidate MovingMan, Forces1D, Ramp, Rotation, Torque into a single codebase. • Better graphics/characters? Beetle will be undesirable for some? ################################################ # Implementation issues ################################ • RotationBody should have internal representation for angular velocity and angular acceleration instead of relying on RotationPlatform's instance? • Better inherent support for different units in charts -ControlGraphSeries should have an optional transform object. ################################################ # Interview: ################################ • how should update mode be set based on typing in text boxes? -current implementation: final edit decides update mode. -For example, if the user wants theta=1, omega=2, alpha=3 and sets them in that order, it will work properly -however, if she enters theta last, then the update mode will be set to "constant position" and therefore omega and alpha will be set to zero -interview on this • Got rid of abbreviations like (vx) on charts, do we want to show abbreviations (e.g. symbol for theta) for other series? -wait until interviews ################################################ # Postponed: ################################ • Investigate acceleration model from Motion2D simulation as a possibility for replacing the implementation in rotation.