flex2treemap - Project Hosting on Google Code

Earlier this year Aug 03, 2009 r131 (Added embedded fonts support to FlexFontUtil) committed by joshtynjala - Added embedded fonts support to FlexFontUtil Added embedded fonts support to FlexFontUtil Jul 28, 2009 issue 15 (Seems to memory leak) Status changed by joshtynjala - Thanks for letting me know. Closing. Status: Invalid Thanks for letting me know. Closing. Status: Invalid Jul 28, 2009 issue 15 (Seems to memory leak) commented on by mullwaden - Im terribly sorry about this. I was being a dumbass, found the error today and its an issue unrelated to the treemap. Sorry. Im terribly sorry about this. I was being a dumbass, found the error today and its an issue unrelated to the treemap. Sorry. Jul 28, 2009 issue 15 (Seems to memory leak) commented on by mullwaden - Im terribly sorry about this. I was being a dumbass, found the today and its an issue unrelated to the treemap. Sorry. Im terribly sorry about this. I was being a dumbass, found the today and its an issue unrelated to the treemap. Sorry. Jul 23, 2009 issue 15 (Seems to memory leak) reported by mullwaden - What steps will reproduce the problem? 1. Made a treemap that is being updated very frequently. What is the expected output? What do you see instead? After a while the amount of objects in the treemap is more or less constant, they only move around and change size. This should mean memory usage evening out, but instead it keeps on increasing. What steps will reproduce the problem? 1. Made a treemap that is being updated very frequently. What is the expected output? What do you see instead? After a while the amount of objects in the treemap is more or less constant, they only move around and change size. This should mean memory usage evening out, but instead it keeps on increasing. Jun 22, 2009 issue 12 (Branch click) commented on by mayannk.sharma - Hi I have a problem in using treemap control i want to use a back button to display to show a previous node while drill down the treemap. please help me on this how can i do the same Hi I have a problem in using treemap control i want to use a back button to display to show a previous node while drill down the treemap. please help me on this how can i do the same Mar 10, 2009 r130 (Fixed bug where changes to weightField and weightFunction at...) committed by joshtynjala - Fixed bug where changes to weightField and weightFunction at runtime would have no effect. Fixed bug where changes to weightField and weightFunction at runtime would have no effect. Jan 20, 2009 issue 13 (a problem with data providers containing only one item ) Status changed by joshtynjala - Good find. This has been fixed in the SVN repository, and it will appear in the next release. Status: Fixed Good find. This has been fixed in the SVN repository, and it will appear in the next release. Status: Fixed Jan 20, 2009 r129 (Fix for bug where setting dataProvider to an Array with a si...) committed by joshtynjala - Fix for bug where setting dataProvider to an Array with a single item would cause a runtime error. Fix for bug where setting dataProvider to an Array with a single item would cause a runtime error. Jan 20, 2009 issue 14 (Branch zoom event is not being dispatched) Status changed by joshtynjala - Thanks! The bug fix has been committed to the SVN repository, and it will be included in the next release. Status: Fixed Thanks! The bug fix has been committed to the SVN repository, and it will be included in the next release. Status: Fixed Jan 20, 2009 r128 (Fixed bug where branchZoom event wasn't being dispatched.) committed by joshtynjala - Fixed bug where branchZoom event wasn't being dispatched. Fixed bug where branchZoom event wasn't being dispatched. Jan 19, 2009 issue 14 (Branch zoom event is not being dispatched) reported by alexander.turetsky - add a listener for a TreeMapEvent.BRANCH_ZOOM. Try to run the application and you'll see that the listener is not called when a branch is being zoomed. After reading the source of the component, it seems that there is a mistake - the needed event is created, but not dispatched. the code from the end of function branchZoomHandler: var zoomEvent:TreeMapEvent = new TreeMapEvent(TreeMapEvent.BRANCH_ZOOM, renderer); this.zoomChanged = true; this.invalidateProperties(); this.invalidateDisplayList(); add a listener for a TreeMapEvent.BRANCH_ZOOM. Try to run the application and you'll see that the listener is not called when a branch is being zoomed. After reading the source of the component, it seems that there is a mistake - the needed event is created, but not dispatched. the code from the end of function branchZoomHandler: var zoomEvent:TreeMapEvent = new TreeMapEvent(TreeMapEvent.BRANCH_ZOOM, renderer); this.zoomChanged = true; this.invalidateProperties(); this.invalidateDisplayList(); Jan 19, 2009 issue 13 (a problem with data providers containing only one item ) reported by alexander.turetsky - What steps will reproduce the problem? 1. Create a tree map with a data provider containing only one item treeMap.dataProvider = [{label: "first"}]; 2. define a color function treeMap.labelFunction = function (item: Object): String { return item.label; } 3. run the application. The application will throw an exception: Error: Unknown Property: 'label'. at mx.collections::ListCollectionView/http://www.adobe.com/2006/actionscript/flash/proxy::getProperty()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\collections\ListCollectionView.as:694] at <anonymous>()[C:\Source\clean40\DecisionIris 4.0\src\Test.mxml:10] at com.flextoolbox.controls::TreeMap/itemToLabel()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1198] at com.flextoolbox.controls::TreeMap/commitItemProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1862] at com.flextoolbox.controls::TreeMap/commitBranchProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1800] at com.flextoolbox.controls::TreeMap/commitProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1475] at mx.core::UIComponent/validateProperties()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:5807] at mx.managers::LayoutManager/validateProperties()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\managers\LayoutManager.as:539] at mx.managers::LayoutManager/doPhasedInstantiation()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\managers\LayoutManager.as:689] at Function/http://adobe.com/AS3/2006/builtin::apply() at mx.core::UIComponent/callLaterDispatcher2()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:8628] at mx.core::UIComponent/callLaterDispatcher()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:8568] When running the application in debugger, I can see that the entire data provider (an ArrayCollection containing only one object) is passed to the labelFunction The same example works fine when the data provider contains 2 items: treeMap.dataProvider = [{label: "first"}, {label: "second"}]; What steps will reproduce the problem? 1. Create a tree map with a data provider containing only one item treeMap.dataProvider = [{label: "first"}]; 2. define a color function treeMap.labelFunction = function (item: Object): String { return item.label; } 3. run the application. The application will throw an exception: Error: Unknown Property: 'label'. at mx.collections::ListCollectionView/http://www.adobe.com/2006/actionscript/flash/proxy::getProperty()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\collections\ListCollectionView.as:694] at <anonymous>()[C:\Source\clean40\DecisionIris 4.0\src\Test.mxml:10] at com.flextoolbox.controls::TreeMap/itemToLabel()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1198] at com.flextoolbox.controls::TreeMap/commitItemProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1862] at com.flextoolbox.controls::TreeMap/commitBranchProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1800] at com.flextoolbox.controls::TreeMap/commitProperties()[/Users/josh/Development/ActionScript 3/projects/flextreemap/output/flextreemap/source/com/flextoolbox/controls/TreeMap.as:1475] at mx.core::UIComponent/validateProperties()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:5807] at mx.managers::LayoutManager/validateProperties()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\managers\LayoutManager.as:539] at mx.managers::LayoutManager/doPhasedInstantiation()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\managers\LayoutManager.as:689] at Function/http://adobe.com/AS3/2006/builtin::apply() at mx.core::UIComponent/callLaterDispatcher2()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:8628] at mx.core::UIComponent/callLaterDispatcher()[C:\autobuild\3.2.0\frameworks\projects\framework\src\mx\core\UIComponent.as:8568] When running the application in debugger, I can see that the entire data provider (an ArrayCollection containing only one object) is passed to the labelFunction The same example works fine when the data provider contains 2 items: treeMap.dataProvider = [{label: "first"}, {label: "second"}]; Older Dec 10, 2008 Examples (Examples that demonstrate how to use various features of the...) Wiki page edited by joshtynjala Dec 10, 2008 flextreemap-2.1.0.zip (TreeMap Component for Adobe Flex 2.1.0) file uploaded by joshtynjala - Labels: Featured Type-Archive OpSys-All Labels: Featured Type-Archive OpSys-All Dec 10, 2008 GettingStarted (Introduction to using the TreeMap component in Adobe Flex) Wiki page edited by joshtynjala Dec 10, 2008 issue 11 (Detect when the zoom button is clicked) Status changed by joshtynjala - There is a new branchZoom event on TreeMap in version 2.1.0. It is fired whenever the zoomed branch changes (zoom in or out) Status: Fixed There is a new branchZoom event on TreeMap in version 2.1.0. It is fired whenever the zoomed branch changes (zoom in or out) Status: Fixed Dec 10, 2008 r125 (Fixed bug in IntroToTreeMap example. Added branchZoom event ...) committed by joshtynjala - Fixed bug in IntroToTreeMap example. Added branchZoom event metadata to TreeMap. Fixed bug in IntroToTreeMap example. Added branchZoom event metadata to TreeMap. Dec 10, 2008 GettingStarted (Introduction to using the TreeMap component in Adobe Flex) Wiki page edited by joshtynjala Dec 10, 2008 BuildInstructions Wiki page deleted by joshtynjala Dec 10, 2008 TreeMap Wiki page deleted by joshtynjala Dec 10, 2008 Examples (Examples that demonstrate how to use various features of the...) Wiki page edited by joshtynjala Dec 10, 2008 Examples (Examples that demonstrate how to use various features of the...) Wiki page edited by joshtynjala Dec 10, 2008 Examples (Examples that demonstrate how to use various features of the...) Wiki page edited by joshtynjala Dec 10, 2008 r118 (Updated examples for 2.1.0. Updated release notes for 2.1.0....) committed by joshtynjala - Updated examples for 2.1.0. Updated release notes for 2.1.0. More documentation tweaks. Updated examples for 2.1.0. Updated release notes for 2.1.0. More documentation tweaks. Dec 10, 2008 r117 (Added separate branch label and datatip properties. Tweaked ...) committed by joshtynjala - Added separate branch label and datatip properties. Tweaked the leaf renderer skin. Added separate branch label and datatip properties. Tweaked the leaf renderer skin. Dec 09, 2008 r116 (Fixed bug where 0 total weight would break squarify layout. ...) committed by joshtynjala - Fixed bug where 0 total weight would break squarify layout. Tweaked build. Fixed bug where 0 total weight would break squarify layout. Tweaked build. Dec 09, 2008 r115 (Fixed disabled behavior. Tweaked the default leaf renderer m...) committed by joshtynjala - Fixed disabled behavior. Tweaked the default leaf renderer mouse states. Fixed disabled behavior. Tweaked the default leaf renderer mouse states. Dec 09, 2008 r114 (Fixed bug in Squarify layout. Tweaked the default leaf rende...) committed by joshtynjala - Fixed bug in Squarify layout. Tweaked the default leaf renderer skin (now different than ButtonSkin to better emphasize selection). Minor documentation tweaks. Fixed bug in Squarify layout. Tweaked the default leaf renderer skin (now different than ButtonSkin to better emphasize selection). Minor documentation tweaks. Dec 09, 2008 r113 (Updated missing documentation. Tweaked build to closer match...) committed by joshtynjala - Updated missing documentation. Tweaked build to closer match the new flexwires style. Updated missing documentation. Tweaked build to closer match the new flexwires style. Dec 09, 2008 r112 (Fixes to new Squarify layout. Can now handle dimensions of z...) committed by joshtynjala - Fixes to new Squarify layout. Can now handle dimensions of zero. Documentation updates. Changed TreeMapLayoutEvent to TreeMapBranchEvent and tweaked event types. TreeMap now dispatched branchZoom event when zoom changes. Fixes to new Squarify layout. Can now handle dimensions of zero. Documentation updates. Changed TreeMapLayoutEvent to TreeMapBranchEvent and tweaked event types. TreeMap now dispatched branchZoom event when zoom changes. Dec 08, 2008 issue 8 (How do I style the TreeMap itself?) Status changed by joshtynjala - Status: Invalid Status: Invalid Dec 08, 2008 r111 (Refactored the default leaf renderer to use button skins (wi...) committed by joshtynjala - Refactored the default leaf renderer to use button skins (without an actual Button as a style). Speed increase. Minor documentation updates. Refactored the default leaf renderer to use button skins (without an actual Button as a style). Speed increase. Minor documentation updates. Dec 08, 2008 r110 (Performance improvements. Thanks Flex profiler! Some docs up...) committed by joshtynjala - Performance improvements. Thanks Flex profiler! Some docs updates. Changed getters and setters in BaseTreeMapData and subclasses to variables. Not needed. Performance improvements. Thanks Flex profiler! Some docs updates. Changed getters and setters in BaseTreeMapData and subclasses to variables. Not needed. Dec 08, 2008 r109 (Created new non-recursive SquarifyLayout class. Renamed the ...) committed by joshtynjala - Created new non-recursive SquarifyLayout class. Renamed the old SquarifyLayout to RecursiveSquarifyLayout. Tweaked SliceAndDiceLayout to use regular Arrays instead of ArrayCollection. Minor documentation updates. Created new non-recursive SquarifyLayout class. Renamed the old SquarifyLayout to RecursiveSquarifyLayout. Tweaked SliceAndDiceLayout to use regular Arrays instead of ArrayCollection. Minor documentation updates. Oct 17, 2008 issue 12 (Branch click) commented on by wieland75 - Sorry, what I mean by "when a branch is zoomed" is when we roll over a branch and that this one is in pop up mode. Thank you Sorry, what I mean by "when a branch is zoomed" is when we roll over a branch and that this one is in pop up mode. Thank you Oct 17, 2008 issue 12 (Branch click) reported by wieland75 - Hello, Is there a way to have a branch click event ? We also need it when a branch is zoomed and then clicked. Thank you Hello, Is there a way to have a branch click event ? We also need it when a branch is zoomed and then clicked. Thank you Sep 21, 2008 issue 11 (Detect when the zoom button is clicked) reported by halldorf - What steps will reproduce the problem? 1. 2. 3. What is the expected output? What do you see instead? What version of the product are you using? On what operating system? Please provide any additional information below. What steps will reproduce the problem? 1. 2. 3. What is the expected output? What do you see instead? What version of the product are you using? On what operating system? Please provide any additional information below. Sep 09, 2008 r108 (Improved performance by using variables rather than getters ...) committed by joshtynjala - Improved performance by using variables rather than getters in some places. Fixed fontSizeMode style bug in lite leaf renderer. Improved performance by using variables rather than getters in some places. Fixed fontSizeMode style bug in lite leaf renderer. Aug 07, 2008 issue 9 (leafclick needs to be clicked twice) Status changed by joshtynjala - This is the proper behavior. Use the "change" event instead of "leafClick" to listen for when the selectedItem changes. The "leafClick" event is fired first, then the selected item changes, and finally the "change" event is fired. Status: Invalid This is the proper behavior. Use the "change" event instead of "leafClick" to listen for when the selectedItem changes. The "leafClick" event is fired first, then the selected item changes, and finally the "change" event is fired. Status: Invalid Aug 07, 2008 issue 8 (How do I style the TreeMap itself?) commented on by joshtynjala - You should style the branch renderer rather than the treemap itself to change the background color. Even the root of the treemap is drawn with a branch renderer. Try this: TreeMap { branchStyleName: "branchStyle"; } .branchStyle { backgroundColor: #ff0000; } You should style the branch renderer rather than the treemap itself to change the background color. Even the root of the treemap is drawn with a branch renderer. Try this: TreeMap { branchStyleName: "branchStyle"; } .branchStyle { backgroundColor: #ff0000; } Aug 07, 2008 issue 10 (Large dataset treemaps run very slowly on Mac's) changed by joshtynjala - Flash Player on Macs is known to be a bit slower. It was worse with PPC Macs, but I don't think they've improved it on Intel Macs to be as fast as Windows yet. Since I know a lot of engineers at Adobe have moved to Macs in the last couple of years, I hope they'll be more likely to improve this situation over the next couple of versions, but obviously, that doesn't help now. To be honest, there's not much I can do here, but let me explain because it's an interesting story of how Flash Player is being pushed to its limits. I made a lot of optimizations for 2.x, and that's why you noticed that this version is a bit faster than 1.x. However, the bottleneck is definitely Flash Player. I've run it through the profiler on many occasions. Even when I can make Flash Player stall for a second or more between redraws with massive data sets, it tells me my code is running fast. I concluded that the slowdown is happening during Flash Player's render phase. The best I can do is limit the number of display list manipulations, like adding and removing children. If you look at the code, you can see the hoops I've jumped through to make it faster in 2.x. All the display objects are direct children of the root treemap so that the branches can all reuse the old item renderers when data changes. Previously, each branch was a separate treemap that had its own cache of renderers, but now they're shared so that if the branching structure changes, it won't lose a ton of perfectly good renderers. I try to ensure that the same renderers keep the same data if possible so that they don't need to redraw themselves as often. Every little bit helped, but I'm hitting the limits of what I can do. Additionally, I've found that when Flash Player needs to draw many display objects that overlap in one location, it slows down severely when other display objects are moved to that location (simple tests proved it by dragging items on top of each other with the mouse, but the treemap component's layout algorithm positioning item renderers pretty much does the same thing). At some point, I want to build an ultra-light treemap component (separate or an extension to this component) without the concept of item renderers. This special treemap will draw its items directly on its graphics object as simple non-skinned shapes. Skins make the number of visible display objects skyrocket. This will bring the number of active and overlapping display objects down to a level that I hope Flash Player will be able to manage much more easily. It won't be as pretty as the fully skinned version, but this lighter treemap should work well for larger datasets such as the ones you're trying to display. Sorry I can't help you right now. Unfortunately, I don't have an ETA on the lighter treemap. Status: Accepted Labels: Priority-Low Performance −Priority-Medium Flash Player on Macs is known to be a bit slower. It was worse with PPC Macs, but I don't think they've improved it on Intel Macs to be as fast as Windows yet. Since I know a lot of engineers at Adobe have moved to Macs in the last couple of years, I hope they'll be more likely to improve this situation over the next couple of versions, but obviously, that doesn't help now. To be honest, there's not much I can do here, but let me explain because it's an interesting story of how Flash Player is being pushed to its limits. I made a lot of optimizations for 2.x, and that's why you noticed that this version is a bit faster than 1.x. However, the bottleneck is definitely Flash Player. I've run it through the profiler on many occasions. Even when I can make Flash Player stall for a second or more between redraws with massive data sets, it tells me my code is running fast. I concluded that the slowdown is happening during Flash Player's render phase. The best I can do is limit the number of display list manipulations, like adding and removing children. If you look at the code, you can see the hoops I've jumped through to make it faster in 2.x. All the display objects are direct children of the root treemap so that the branches can all reuse the old item renderers when data changes. Previously, each branch was a separate treemap that had its own cache of renderers, but now they're shared so that if the branching structure changes, it won't lose a ton of perfectly good renderers. I try to ensure that the same renderers keep the same data if possible so that they don't need to redraw themselves as often. Every little bit helped, but I'm hitting the limits of what I can do. Additionally, I've found that when Flash Player needs to draw many display objects that overlap in one location, it slows down severely when other display objects are moved to that location (simple tests proved it by dragging items on top of each other with the mouse, but the treemap component's layout algorithm positioning item renderers pretty much does the same thing). At some point, I want to build an ultra-light treemap component (separate or an extension to this component) without the concept of item renderers. This special treemap will draw its items directly on its graphics object as simple non-skinned shapes. Skins make the number of visible display objects skyrocket. This will bring the number of active and overlapping display objects down to a level that I hope Flash Player will be able to manage much more easily. It won't be as pretty as the fully skinned version, but this lighter treemap should work well for larger datasets such as the ones you're trying to display. Sorry I can't help you right now. Unfortunately, I don't have an ETA on the lighter treemap. Status: Accepted Labels: Priority-Low Performance −Priority-Medium Aug 07, 2008 issue 10 (Large dataset treemaps run very slowly on Mac's) reported by narced133 - What steps will reproduce the problem? 1. Create a large (430 entries across 3 levels) XMLList, XMLCollection or ArrayCollection dataset with a treemap 2. Compile with either the latest 1.x or 2.0 beta libraries 3. Run on a Mac in Safari, Opera, IE or Firefox What is the expected output? What do you see instead? The load and response times are extremely slow to the point of being unusable. Zooming in is relatively responsive in 2.0 but not 1.x. Selection, zooming out and modifying color and weighting schemes are slow in both versions. What version of the product are you using? On what operating system? Problems appear in both 1.x and 2.0 on various browsers running under OS X on several relatively new MacBooks and MacBook Pros. Please provide any additional information below. The map works perfectly and looks great PC's though it had the same slow loading problem once when embedded in a Powerpoint. Reloading the slideshow fixed the problem. What steps will reproduce the problem? 1. Create a large (430 entries across 3 levels) XMLList, XMLCollection or ArrayCollection dataset with a treemap 2. Compile with either the latest 1.x or 2.0 beta libraries 3. Run on a Mac in Safari, Opera, IE or Firefox What is the expected output? What do you see instead? The load and response times are extremely slow to the point of being unusable. Zooming in is relatively responsive in 2.0 but not 1.x. Selection, zooming out and modifying color and weighting schemes are slow in both versions. What version of the product are you using? On what operating system? Problems appear in both 1.x and 2.0 on various browsers running under OS X on several relatively new MacBooks and MacBook Pros. Please provide any additional information below. The map works perfectly and looks great PC's though it had the same slow loading problem once when embedded in a Powerpoint. Reloading the slideshow fixed the problem. Aug 05, 2008 issue 9 (leafclick needs to be clicked twice) reported by koen.wou...@eu.sony.com - What steps will reproduce the problem? 1. create a treemap 2. selectable = true 3. leafclick = call a function to do something with the selected leaf (eg label.text = treemap.selectedItem.@id.toString()) What is the expected output? What do you see instead? expected : script would run on single click instead : leaf needs to be clicked twice before the id is passed to the function What version of the product are you using? On what operating system? TreeMap 2.0Beta - Windows XP Please provide any additional information below. What steps will reproduce the problem? 1. create a treemap 2. selectable = true 3. leafclick = call a function to do something with the selected leaf (eg label.text = treemap.selectedItem.@id.toString()) What is the expected output? What do you see instead? expected : script would run on single click instead : leaf needs to be clicked twice before the id is passed to the function What version of the product are you using? On what operating system? TreeMap 2.0Beta - Windows XP Please provide any additional information below. Jul 28, 2008 issue 8 (How do I style the TreeMap itself?) commented on by haiming227 - How do import flexbuilder3? How do import flexbuilder3?

issue 10 (Large dataset treemaps run very slowly on Mac's) changed by joshtynjala - Flash Player on Macs is known to be a bit slower. It was worse with PPC Macs, but I don't think they've improved it on Intel Macs to be as fast as Windows yet. Since I know a lot of engineers at Adobe have moved to Macs in the last couple of years, I hope they'll be more likely to improve this situation over the next couple of versions, but obviously, that doesn't help now. To be honest, there's not much I can do here, but let me explain because it's an interesting story of how Flash Player is being pushed to its limits. I made a lot of optimizations for 2.x, and that's why you noticed that this version is a bit faster than 1.x. However, the bottleneck is definitely Flash Player. I've run it through the profiler on many occasions. Even when I can make Flash Player stall for a second or more between redraws with massive data sets, it tells me my code is running fast. I concluded that the slowdown is happening during Flash Player's render phase. The best I can do is limit the number of display list manipulations, like adding and removing children. If you look at the code, you can see the hoops I've jumped through to make it faster in 2.x. All the display objects are direct children of the root treemap so that the branches can all reuse the old item renderers when data changes. Previously, each branch was a separate treemap that had its own cache of renderers, but now they're shared so that if the branching structure changes, it won't lose a ton of perfectly good renderers. I try to ensure that the same renderers keep the same data if possible so that they don't need to redraw themselves as often. Every little bit helped, but I'm hitting the limits of what I can do. Additionally, I've found that when Flash Player needs to draw many display objects that overlap in one location, it slows down severely when other display objects are moved to that location (simple tests proved it by dragging items on top of each other with the mouse, but the treemap component's layout algorithm positioning item renderers pretty much does the same thing). At some point, I want to build an ultra-light treemap component (separate or an extension to this component) without the concept of item renderers. This special treemap will draw its items directly on its graphics object as simple non-skinned shapes. Skins make the number of visible display objects skyrocket. This will bring the number of active and overlapping display objects down to a level that I hope Flash Player will be able to manage much more easily. It won't be as pretty as the fully skinned version, but this lighter treemap should work well for larger datasets such as the ones you're trying to display. Sorry I can't help you right now. Unfortunately, I don't have an ETA on the lighter treemap.

Status: Accepted

Labels: Priority-Low Performance −Priority-Medium

Flash Player on Macs is known to be a bit slower. It was worse with PPC Macs, but I don't think they've improved it on Intel Macs to be as fast as Windows yet. Since I know a lot of engineers at Adobe have moved to Macs in the last couple of years, I hope they'll be more likely to improve this situation over the next couple of versions, but obviously, that doesn't help now. To be honest, there's not much I can do here, but let me explain because it's an interesting story of how Flash Player is being pushed to its limits. I made a lot of optimizations for 2.x, and that's why you noticed that this version is a bit faster than 1.x. However, the bottleneck is definitely Flash Player. I've run it through the profiler on many occasions. Even when I can make Flash Player stall for a second or more between redraws with massive data sets, it tells me my code is running fast. I concluded that the slowdown is happening during Flash Player's render phase. The best I can do is limit the number of display list manipulations, like adding and removing children. If you look at the code, you can see the hoops I've jumped through to make it faster in 2.x. All the display objects are direct children of the root treemap so that the branches can all reuse the old item renderers when data changes. Previously, each branch was a separate treemap that had its own cache of renderers, but now they're shared so that if the branching structure changes, it won't lose a ton of perfectly good renderers. I try to ensure that the same renderers keep the same data if possible so that they don't need to redraw themselves as often. Every little bit helped, but I'm hitting the limits of what I can do. Additionally, I've found that when Flash Player needs to draw many display objects that overlap in one location, it slows down severely when other display objects are moved to that location (simple tests proved it by dragging items on top of each other with the mouse, but the treemap component's layout algorithm positioning item renderers pretty much does the same thing). At some point, I want to build an ultra-light treemap component (separate or an extension to this component) without the concept of item renderers. This special treemap will draw its items directly on its graphics object as simple non-skinned shapes. Skins make the number of visible display objects skyrocket. This will bring the number of active and overlapping display objects down to a level that I hope Flash Player will be able to manage much more easily. It won't be as pretty as the fully skinned version, but this lighter treemap should work well for larger datasets such as the ones you're trying to display. Sorry I can't help you right now. Unfortunately, I don't have an ETA on the lighter treemap.

Status: Accepted

Labels: Priority-Low Performance −Priority-Medium

Earlier this year

Older