================================================ Known Issues That Would Need Serious Work to Fix ================================================ o Cannot have multiple editors connecting to the AIBO, as all of the editors have keep-alive connections and the AIBO only has one port open for listening to these connections. In other words, it isn't effective to have multiple editor window tabs (files) open at one time. o On hover of message icon, tooltip with value of message should show after zero delay (currently there is a delay). Tooltip draw speed is based on the OS delay for mouse hover. o The algorithm for finding the current selection in the storyboard view is suboptimal. A hB-tree would be required to reduce the complexity from O(n) to about O(log n) where n is the number of objects currently in the storyboard view. Currently, the selection algorithm runs in O( n log n ) worst case, but is optimized to be O(log n) when the selection cursor is towards the end of the storyboard (the application uses the most CPU time when new nodes are being added). oooooooooooooooooooooooooooooooo o Unreproducable Reported Bugs o oooooooooooooooooooooooooooooooo 2009.5.31 (opd) - If I delete the "webcam" node, boths its transitions go away, as they should. If I then right click and select "Add State Node", I can bring the node back. But if I then right click and select "Add Transition", I can only bring back one of its transitions, because it doesn't think the other one has been deleted. However, if I right click and select Generate Layout, it brings back the other one. - Maybe the right heuristic is to say that the width of a box in the execution trace should be no less than its width in the layout. (It can be greater if the state is active for a long time, but it can't be less.) - Java occasionally displays the following error message: Exception in thread "Thread-2" org.eclipse.swt.SWTException: Invalid thread access at org.eclipse.swt.SWT.error(SWT.java:3777) at org.eclipse.swt.SWT.error(SWT.java:3695) at org.eclipse.swt.SWT.error(SWT.java:3666) at org.eclipse.swt.widgets.Display.error(Display.java:1142) at org.eclipse.swt.widgets.Display.checkDevice(Display.java:704) at org.eclipse.swt.widgets.Display.getActiveShell(Display.java:1328) at org.tekkotsu.ui.editor.netcode.ModelConnector.connect(ModelConnector.java:67) at org.tekkotsu.ui.editor.netcode.ModelConnector$ConnectorThread._changeCommand(ModelConnector.java:306) at org.tekkotsu.ui.editor.netcode.ModelConnector$ConnectorThread.run(ModelConnector.java:218) 2009.1.1 (opd) - Message icons (little books) in the Storyboard pane are sometimes buried beneath the node instead of appearing on top of it. But with later redraws they eventually end up on top of the node. 2008.12.29 r66 (opd) [5 mins] - If you run the Bandit State Machine (it's under Behaviors/Demos) and download the model, you'll see that some transitions are missing. 2008.05.30 (opd) r14 [15 mins] - Hexagon shapes cannot display in the proper color in the layout window. Other shapes display correctly. - When you first load a layout and model, or when you click in an empty area of the Layout pane, the Properties tab shows a "Model source..." line. Clicking on this line produces: Unhandled event loop exception Reason: Assertion failed: ============ Change Log ============ 2008.05.16 (opd) r1 [2 hours] - Built OPEN_R_SDK - Built Tekkotsu bleeding source (someone broke the build and I had to "fix" it) - Setup private SVN work repository - Ran unchanged Storyboard Plugin 2008.05.17 (opd) r5 [2 hours] - Updated plugin to Eclipse Europa 3.3.2 - Fixed all errors as reported by Eclipse 2008.05.17 (opd) r6 [2 hours] - Updated code to use Java 5/6 generics - Fixed all errors/warnings as reported by Eclipse 2008.05.17 (opd) r7 [2 hours] - Fixed all errors/warnings as reported by FindBugs 2008.05.23 (opd) r12 [3 hours] - Removed all deprecated function calls - Updated plugin.xml - FIXED: & Change the window name from "Tekkotsu Viewer" to "Storyboard Tool". 2008.05.26 (opd) r14 [2 hours] - FIXED: & Can't tell Pause mode from un-Pause mode because the button doesn't change symbol or color. & The code sometimes prints "Rectangle" messages, e.g.: Rectangle(0, 53, 129637, 237) Rectangle(0, 53, 130927, 447) + If you click on a node in the Storybboard pane, the program should automatically switch from the Properties tab to the Runtime View tab. + If you click on a node in the Layout pane, the program should automatically switch from the Runtime View tab to the Properties tab. - REDIAGNOSED: + Clicking on the "Open a layout" (not the "Open Model") icon and then selecting Cancel generates a Null Pointer exception. 2008.05.30 (opd) r15 [15 mins] - FIXED: + Clicking on the "Open a layout" (not the "Open Model") icon and then selecting Cancel generates a Null Pointer exception. 2008.05.30 (opd) r17 [1.5 hours] - Apparently some of the menus were lost from previous changes. Added the menus back. - Known bugs sorted by related issue. - FIXED: + If you close the Layout or Storyboard pane and then open a new pane, something is screwed up internally and you can no longer load an execution trace. 2008.05.30 (opd) r18 [4 hours] - Redid layout of Editor View to be more intuitive. - FIXED: + Replace Monitor/Layout tabs with a big fat arrow pointing up or down. 2008.06.08-13 (opd) [5 hours] - Miscellaneous GUI code cleaning. - Researching more about available GUI components and their specifications. 2008.06.15 r19 (opd) [1 hour] FIXED: - When you start the program, it should pop up a dialog offering to either load an existing model/layout or let you download a new model. (don't need this behavior any more). - New layout pane should be generated automatically on startup. - Note to self: You can't do anything with the workbench until after it is initialized by eclipse XD. 2008.06.15 r20 (opd) [2 hours] - Identified places in code where need to set dirty flag for the StateMachineEditor, but need to get a hold of a reference to the editor. Might be able to get it via PlatformUI? - Removed places where the the dirty flag for the StateMachineEditor was set but shouldn't have been. - Note to self: This part of the code is spagetti without documentation -_-. Set breakpoints on Debugger.printDebug to find stuff. 2008.06.18 r21 (opd) [30 minutes] FIXED: + If you click on the Layout pane, the "Save" icon ungrays even though you haven't made any changes to the layout. + In the Layout tab, moving a node or link does not un-dim the Save button. 2008.06.19 r22 (opd) [30 minutes] FIXED: + Relative file names don't work, so layout file can't properly reference model file. 2008.06.20 r23 (opd) [1 hour] FIXED: + Directory choices in file browser should be sticky. o I hate '\r' o ".tekkotsu_settings/settings.ini" gets stored across sessions in whatever directory the program starts up in... which on unix is typically /home// 2008.06.21 r24 (opd) [2 hours] FIXED: + Downloading a model, and then opening a different model, and then trying to download the model again produces errors: java.net.ConnectException: Connection refused o When the editor is disposed we have to close the connection so that the port is open again for any new editors. o You still can't connect over multiple editors. 2008.06.28 r25 (opd) [30 minutes] + Removed the menubar with save/open/load icons. Called the "coolbar" in eclipse rcp speak. Commands are accessible from the file menu. 2008.07 05 r26 (opd) [1 hour] + When you download a model and automatically generate a layout you get an error message pop up saying: "No model loaded, you must pick an existing model or upload a new one." o Had to do with downloading an empty model o Now the plugin generates a warning when this happens 2008.07.06 r26 (opd) [2 hours] FIXED: + Undocumented bug dealing with needing to strip leading and trailing whitespace from node and transition IDs. + If you load a model containing a transition that references a non-existent state, the program spazzes. It should just generate a dummy state and continue. 2008.07.12 r27 (opd) [1 hour] + r26 "fix" from 2008.07.06 actually caused another bug when executing a trace. The real problem was that deep in the code whitespace was being stripped where it shouldn't have been stripped. o Also removed some more stray printlns(). 2008.07.13 r31 (opd) [30 mins] + Reverted broken changes made since r27. + Fixed change log. 2008.07.19 r32 (opd) [2 hours] + JDOM XML "prettyfier" in the original code was killing meaningful whitespace when you saved a file. Problem only presented itself after fixing the bug mentioned on 2008.07.12 2008.07.29 r33 (opd) [30 minutes] + Found more references to JDOM prettyfier :( 2008.07.29 r34 (opd) [2 hours] + Fixed a really bad bug which made it so you couldn't execute new traces once any open editor window was closed. 2008.07.30 r35 (opd) [2 hours] + Updated the main menu bar to be simpler and replaced all icons with text for ease of understanding. + Tested various combinations of button pressing... (both logical and idiotic). o For logical sequences, seems pretty stable. o For idiotic sequences, sometimes will produce weird results, but always recoverable by performing a "New Trace" o Design could probably be improved 2008.07.31 r36 (opd) [15 minutes] + Added back the "preview mode" for the layout window when a trace is executing so that the colors adjust to highlight active states. o Surprisingly easy for once =) 2008.07.31 r37 (opd) [30 minutes] + Fixed sticky paths as much as reasonably possible (see known "hard to fix" bugs at top). 2008.07.31 r38 (opd) [15 minutes] + FIXED - Hitting cancel while loading or saving an execution trace produced an error rather than just a no-op. 2008.07.31 r39 (opd) [1 hour] + Prepared for "stable" build to be shipped off. + Turned off debugging prints. + Tested Deployment and updated README 2008.07.31 r40 (opd) [1 hour] + Updated Plugin to Eclipse Ganymede 3.4 + FIXED (via updating to Ganymede + some other things) o Due to a bug in the Eclipse RCP's interaction with GTK on linux, a "sticky" file path has a few problems. - Option 1: The directory above the saved directory gets opened. - Option 2: The last file opened is "sticky"; if that file has a different extension, then the filter is initially goofed up. o I've opted for the behavior in option 2 as it is closer to the expected behavior. 2008.09.09 (opd) [1.5 hours] - Meeting to discuss new bugs and design changes. 2008.09.20 (opd) [1 hour] - Played around with the stuff I was having trouble debugging before because I didn't know how to get images/messages from the "robot". 2008.09.27 (opd) [1 hour] - Added bugs discussed in meeting to buglist and created a new plan of attack. 2008.10.03 (opd) [3 hours] - Got a feel for the buggy storyboard display code and took notes to be used later when I need to remember what part of the code is doing what (previous coder was poorly organized) 2008.12.18 (opd) [30 mins] - Reaquianted self with tools/code + Remove the port field 2008.12.19 (opd) [1 hour] - Code cleanup + FIXED - Change initial state machine name from "Bandit State Machine" to "" and make the box red until the user types something in it. (surprising how many hoops you jump through to change a color =/) 2008.12.19 r43 (opd) [10 mins] + FIXED - When you highlight the Behavior Name field and it is red, it should turn white and blank out the text. 2008.12.20 (opd) [2 hours] - Trying to make sense of the storyboard display code. =/ 2008.12.21 r45 (opd) [2 hours] - More trying to make sense of storyboard code. - Removed code that didn't do anything + FIXED - Algorithm for deciding whether to use white or black font color based on background color. 2008.12.21 r48 (opd) [30 mins] + FIXED - Application now takes command line arguments. o --host o --behavior <"name of behavior"> 2008.12.21 r49 (opd) [30 mins] + FIXED - When nothing is selected, it should read: "click on a node or link in the layout" (had to hack it) 2008.12.21 r50 (opd) [2 hours] + FIXED - Added --layout command line argument - Fixed resource leak in NewFileAction and OpenLayoutFileAction 2008.12.23 r51 (opd) [20 mins] + Reproduced and found bug in code, left TODO for later. (a message box should show up) - Java is throwing an unhandled exception if the robot can't connect because the user specified the wrong IP address (specific errors) (check port) 2008.12.23 r52 (opd) [1.5 hours] + FIXED o Properties View - Things under Info should be unedittable - Things under Misc should be visually distinct - Misc items shouldn't be under a submenu (get rid of Misc) o Note: Properties View is a built in Eclipse thing, so there isn't much flexibility here. Not fixed exactly as requested. 2008.12.23 r53 (opd) [1 hour] + Cleaned up the way exceptions are reported. TODO: add a debug level flag to command line args. 2008.12.23 r54 (opd) [20 mins] + FIXED - Added a debug level flag to command line args. 2008.12.27 r54 (opd) [15 mins] + FIXED - Doing File > Close All while tracing causes exceptions (just close and stop tracing) 2008.12.27 r55 (opd) [15 mins] - Organized Automatic Layout generator code and factored out anonymous classes into named actions. 2008.12.27 r56 (opd) [4 hours] + FIXED - When the system automatically picks colors for state nodes, it does so by traveling around a circular list so that the selected colors will be equally spaced. However, if there are more than about 6 states, the list repeats, and several unrelated nodes end up with the same color. Can you fix the algorithm to always generate a unique color for each node? - Other miscellaneous organization issues with node creation code. %%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Total so far: 68.15 hours % %%%%%%%%%%%%%%%%%%%%%%%%%%%%% 2008.12.28 r57 (opd) [4 hours] - Failed attempt at allowing font color to be manually overridden and saved to the XML layout files. Had to revert to previous revision in SVN because I broke something completely unrelated somehow. + FIXED (and discovered) - Exceptions thrown and editor state inconsistent and unsavable if a file that doesn't exist is opened (via command line). 2008.12.28 r58 (opd) [5 hours] - Delved deeper into the horrors of undocumented code making calls to the poorly documented GEF library. - Removed a bunch of unnecessary code. + FIXED - Should be able to move transition arrow endpoints around the edge of a node. (because of GEFs LayoutManager this wasn't fixed perfectly) 2008.12.29 r59 (opd) [3 hours] - Realized how horrible the editpart heirarchy is. + FIXED - Editting the shape of an arrow throws an exception and then editting other things is broken. - Changing the width of a line doesn't flag the editor dirty. - Selecting a transition does not bring up its properties. 2008.12.29 r60 (opd) [1 hour] + FIXED - Line color is saved but not properly loaded. 2008.12.29 r61 (opd) [1 hour] + FIXED - Cannot use XML prettifier because whitespace can be in a state or transition id. (need to escape the whitespace). 2008.12.29 r64 (opd) [1 hour] + FIXED - Non-backwards-compatible changes to the model XML file schema now will fall back to the old version if the new version isn't used. 2008.12.29 r65 (opd) [1 hour] - Revisions on multi-headed/tailed transitions. + FIXED - Transition labels show when they shouldn't then go away after some operation (redraw or reload?) - If you delete a multi-tailed transition and then add it again, the transition node has a label like "transition19" that is visibly displayed and won't go away. However, if you switch to the Layout Source pane and then switch back to the Layout pane, the label disappears. (What actually happens is that the label text switches from back to white.) 2008.12.29 r66 (opd) [1 hour] + FIXED - If you delete one arrow of a multi-headed transition, there is no way to add it back, because the Add Transition menu still thinks the transition is present. You must delete the entire transition and then add it back. (had to fix by making the whole thing delete when you delete any part of the multi-transition). 2008.12.31 r67 (opd) [2 hours] + FIXED - --trace command line argument to open a trace file on startup. 2008.12.31 r75 (opd) [5 hours] - Cleaning, refactoring and understanding the Storyboard display code. + FIXED - background and font colors made consistent with editor. 2008.12.31 r75 (opd) [1 hour] + FIXED - In the storyboard pane, when a node has been active for a long time it has a lengthy box. The system tries to center a state node name in the box, but really long nodes might be wider than the current frame so the node name ends up out-of-frame and not visible. Can you change the display code so that node names are always visible when any part of the node is within the current frame? 2008.12.31 r76 (opd) [1 hour] + FIXED - Variable timeline only works sometimes. 2008.12.31 r77 (opd) [1 hour] - Got old samples to work with changes. 2009.1.1 r78 (opd) [2 hours] + FIXED - Image Preview view working properly. 2009.1.1 r79 (opd) [2 hours] - More code cleanup. + FIXED - Mousing on a thumbnail image does not display the image or move the time index cursor. The mouse must be above or below the thumbnail in order to get the cursor to move. 2009.1.1 r81 (opd) [1 hour] - Cannot fix tooltip hover delay. Based on OS's delay for hover. 2009.1.1 r82 (opd) [30 mins] + FIXED - Clicking "New Trace" does not reset the length of the Storyboard timeline, so the scrollbar retains its current size. 2009.1.1 r83 (opd) [2 hours] + FIXED - When storing a model, include the name of the parent state. When loading a model, set the state machine name to this parent state. - A problem with the way relative paths were being stored in the layout xml. 2009.1.1 r84 (opd) [1.5 hours] + FIXED - Set minimum size to button bars so buttons can never be hidden by changing the size of the windows 2009.1.1 r85 (opd) [2 hours] + FIXED - Record/Pause changes + layout of buttons. 2009.1.1 r86 (opd) [1 hour] + FIXED - Running LogTestMachine on the simulator, the Storyboard spazzes each time a transition occurs. Lots of extraneous redraws; shapes drawn in wrong size/place and then redrawn. 2009.1.1 r88 (opd) [2 hours] + FIXED - Dragging the cursor (red line) while a trace is being generated does weird stuff; the active state boxes stretch to follow the cursor. - Miscellaneous GUI changes. 2009.1.1 r91 (opd) [30 mins] + FIXED - Force complete storyboard redraw when something changes in the editor. 2009.1.2 r92 (opd) [1 hour] + FIXED - O(n) algorithm for updating the storyboard improved to O(1). 2009.1.2 r96 (opd) [3 hours] + FIXED - Timeline ticks are weird. 2009.1.2 r98 (opd) [1 hour] - Miscellaneous polishing and tweaks. 2009.1.2 r109 (opd) [4 hours] - Failed at hacking together a proper scrolling storyboard because there were too many things trying to get it to work already and messing with any one of them screwed everything up. - Started reorganizing the code. + FIXED - If the scroll bar is hard against the right edge, then even if a state transition occurs, the timeline will not grow. The scrollbar must be slid leftward a bit, and then a state transition must occur, in order for the timeline to grow. - Currently, if the AIBO remains in one state as the time index advances, the index will march off the right end of the Storyboard pane and disappear. The scroll bar does not advance or grow; there is no way to scroll to the right until after some state transition has occurred. 2009.1.2 r121 (opd) [2 hours] - Cleaned up editor part activation code. 2009.1.3 r136 (opd) [6 hours] - More clean up of code, improving selection algorithms and making the storyboard code consistent to make it easier to deal with scrolling issues. 2009.1.4 r156 (opd) [5 hours] - While continuing to clean up the code, discovered a race condition in the networking code that grabs data from the robot. Drastically cleaned up the net code and fixed the race condition. Discovered that the connection's IO was not blocking and that the updates to the storyboard tool could be more real time. 2009.1.4 r156 (opd) [2 hours] - Net code completely fixed. Storyboard now blocks waiting for input to read rather than alternating between trying to read and sleeping for one second. 2009.1.4 r157 (opd) [1 hour] - Scroll bars in Storyboard viewer now shrink correctly. 2009.1.4 r158 (opd) [30 mins] + FIXED - RuntimeView: the times should be cut off after some number of decimal places 2009.1.4 r162 (opd) [3 hours] + FIXED - Horizontal scroll bar in Storyboard pane needs to advance with time. 2009.1.4 r164 (opd) [2 hours] + FIXED - Please verify that the event icons work correctly in your version. To test this I think you need to go into the Event Logger and turn on monitoring for various event generators, like buttonEGID and audioEGID. I'm not sure what events it currently recognizes and what the list of icons is, so it would be good if you could point me to that info. We'll be adding new event types later, so I should learn how to do that. You can simulate buttons press events by typing !post buttonEGID HeadBut A in the ControllerGUI's "Send Input" window. If HeadBut doesn't work then use a small integer like 2 instead. o This is the algorithm for selecting an icon for a log event: - All icons must be in PNG format (and have a ".png" extension) - All icons must be located in the directory src/org/tekkotsu/ui/storyboard/icons/ - The filename of the image is selected by choosing the first one of the following that exist: 1) The value of the "icon" attribute in the XML for the event 2a) The value of the "egid" attribute in the XML for the "log" event 2b) The value of the "sid" attribute in the XML for the "userlog" event 3) "__fallback__.png" 2009.1.5 r167 (opd) [3 hours] - Improved the algorithm that selects and displays log items. - Fixed a bug that prevented both log items and state/transitions from simultaneously showing in the runtime view. 2009.1.5 r175 (opd) [3 hours] + FIXED - When drawing the trace, take the firing time of the transition to also be the stop time of the sources and the start time of the destinations. That way everything lines up nicely. If the user puts the cursor over a state he can see the actual start and stop times for that state, so what's really going on is not totally hidden from him. But the graphical execution trace will look sensible. 2009.1.5 r177 (opd) [2 hours] + FIXED - Unfinished nodes do not display. - Weird font alignment display glitches while scrolling in small increments. 2009.1.5 r180 (opd) [4 hours] + FIXED o Bendpoints added to transitions are only kept for the current session, they are not saved to the XML files. This doesn't seem worth the effort to fix because it would require a complete rewrite of the multi-head/tail connection code. 2009.1.6 r181 (opd) [4 hours] o CVS Update - Also, I changed the format for displaying Transition names; instead of foo--bar-->baz it's not foo>==bar==>baz to match the new state machine notation. Not sure if this will affect your code at all, but to be safe, do a CVS udpate and recompile Tekkotsu. + FIXED - Miscellaneous bugs caused by slight changes in the way XML is sent to the storyboard tool from the robot. 2009.1.21 (opd) [5 hours] o Fixed Plugin Binary in Student Lab o Fixed other misc bugs. o TODO: add new bugs to buglist to fix. 2009.2.22 (opd) [5 hours] - Massive cleanup of file IO + FIXED - Storyboard just failed for one of my students because it's trying to write the workspace files in the course directory where the tool resides, instead of in the student's home directory. I've patched this by giving a -data switch, but now it's complaining about to not being able to write the temporary layout file. It runs, though, so we're okay for today. Can you fix it to put the workspace files in the home directory, and not write out dummy noname files on initial startup, which seems pointless? > Each editor must be linked to a file in eclipse. Upon creating a new > layout, I could prompt the user for a file to save to... Yes, but that would be distracting. Instead, just silently open a dummy file in /tmp that contains the pid as part of the name, e.g., /tmp/storyboard-temp-12358.tsl. + Since Java cannot get pids, I used a random long. + The temporary files get removed when the storyboard exits. 2009.2.23 (opd) [2 hours] - Further cleanup of FileIO + FIXED - Right now, when there are multiple layout tabs, the filenames displayed in the tab are truncated on the right. If I have long pathnames, like /afs/cs/academic/class/15494-s09/examples/foo.tsl and .../bar.tsl, then I can't tell them apart because of the truncation. It would be better if the file paths in the tab were truncated on the left, so that the tabs read ...examples/foo.tsl and ...examples/bar.tsl. - When the user clicks on Save Layout, if we know the behavior name, use that as the initial guess for the file name when we pop up the file selection dialog box. - When people download a model and start editing it, at some point the behavior name is cleared and the box goes back to the bright red "type your behavior name" thing. This should never happen. - Right now we have a Load Layout button but an Open Layout menu item in the File menu. These should be consistent. I guess we should go with Open Layout for both? 2009.2.23 (opd) [15 mins] + FIXED - I run "Storyboard -h localhost -b LogTestMachine". Then I manually change the host to "chiara4e" and hit Reconnect. So far, so good. Then I do "Load Layout" and it resets the host to "localhost". This should never happen. 2009.2.23 (opd) [3 hours] + FIXED - Load working again (mostly) - Load completely working again. - Combine model xml and layout xml into one file. - As soon as we can guess an appropriate name for the file, change it. - We may have already covered this: when I start up the Storyboard and do "Load Layout", it asks me if I want to save the existing noname.tsl file. This is a distraction. It should recognize that this is an empty, dummy layout, and just silently ditch it. 2009.2.23 (opd) [1 hour] + FIXED - On event receive, sync with local time %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Total so far: 156.15 hours % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 2009.5.25 (opd) [30 mins] + FIXED - Move "Reconnect" button to right of "Load Layout" - Added a dummy clear layout button. 2009.5.25 (opd) [2 hours] + FIXED - Can you add a "Clear Layout" button to erase all the nodes and links? 2009.5.25 (opd) [2 hours] + FIXED - When generating a layout, if two transitions going in opposite directions would overlap and look like a double-headed transition, generate bendpoints so that the transitions look distinct 2009.5.31 (opd) [2 hours] - Briefly looked into fixing a few things until I realized they would be quite a bit of effort. + FIXED - We're currently graying unselected transitions in the execution trace. This is fine when all the transitions are black. But when we use color transitions, the graying combined with the thinness of the lines makes it hard to see the color. This seems to be a little worse in the new version of the Storyboard compared to the old. Here's a screen capture from the old version; I cannot get that bright a shade of pink/red in the new version no matter what I do. How about if we didn't gray out the transition lines, but instead indicated a transition was active by making the line thicker? This is good for another reason: right now the cursor is exactly as thick as the transition, so when we select the transition it becomes obscured. Making it thicker would ensure that we can always see it. 2009.5.31 (opd) [1 hour] + FIXED - The thumbnails in the execution trace are permanently dimmed. It would be more intuitive to un-dim the thumbnail at the same time as you display the larger image in the Image Preview pane (which may not be visible, depending on what the user is doing.) %%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Total so far: 163.65 hours % %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%