as3reflect - Project Hosting on Google Code

Project Home

Older Apr 09, 2009 issue 7 (Potential solution for classUtils::newInstance(clazz,args)) Status changed by martino.piccinato - Ok, this is very interesting in general but the issue tracker is not the right place to discuss this unless you have some more specific patch. I'm closing this issue. Status: Fixed Ok, this is very interesting in general but the issue tracker is not the right place to discuss this unless you have some more specific patch. I'm closing this issue. Status: Fixed Apr 09, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) Status changed by martino.piccinato - Now we have as3commons logging. I have used it to log this issue in as3reflection. Default logger is still trace but you can change it with whatever other logger you want and/or filter it. Status: Fixed Now we have as3commons logging. I have used it to log this issue in as3reflection. Default logger is still trace but you can change it with whatever other logger you want and/or filter it. Status: Fixed Apr 09, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) Status changed by martino.piccinato - Ok, I've fixed the bug and added a test for it, checking that the constructor is called just once even if there a Type.forInstance in the class constructor itself. I'm also going to add some debug logging in the catch part of the try/catch constructor workaround blog . It was really a naive assumption on my side to think this would never cause problems. Never, ever ignore exceptions... I've credited you for the patch, got the name from the code. Thanks for all and sorry for the confusion :-) Status: Fixed Ok, I've fixed the bug and added a test for it, checking that the constructor is called just once even if there a Type.forInstance in the class constructor itself. I'm also going to add some debug logging in the catch part of the try/catch constructor workaround blog . It was really a naive assumption on my side to think this would never cause problems. Never, ever ignore exceptions... I've credited you for the patch, got the name from the code. Thanks for all and sorry for the confusion :-) Status: Fixed Apr 08, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) commented on by goo...@anycpu.com - I'm not sure how you would write an automated unit test that produces a partially filled Type instance. The infinite recursion happens until a stack overflow occurs. The stack overflow error is caught by a try ... catch block. Type doesn't count the number of recursions, therefore there is currently nothing observable by a unit test. You need to use the debugger and step through the forInstance() method to see the recursion issue. I'm not sure how you would write an automated unit test that produces a partially filled Type instance. The infinite recursion happens until a stack overflow occurs. The stack overflow error is caught by a try ... catch block. Type doesn't count the number of recursions, therefore there is currently nothing observable by a unit test. You need to use the debugger and step through the forInstance() method to see the recursion issue. Apr 08, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) commented on by martino.piccinato - Sorry for not spotting the class, something must have happened with my patch program. The problem is that I can't reproduce the reported behaviour with this test, all tests just run fine without your change. Could this be platform dependent? The statements swap seems reasonable and I can definitely do it but I'd like to see a test fail first and pass after to be sure we actually fixed this, any idea? Sorry for not spotting the class, something must have happened with my patch program. The problem is that I can't reproduce the reported behaviour with this test, all tests just run fine without your change. Could this be platform dependent? The statements swap seems reasonable and I can definitely do it but I'd like to see a test fail first and pass after to be sure we actually fixed this, any idea? Apr 08, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) commented on by martino.piccinato - Sorry for not spotting the class something must have happened with my patch program. The problem is that I can't reproduce the reported behaviour with this test, all tests just run fine. The statements swap seems reasonable and I can definitely do it but I'd like to see a test fail first and pass after, any idea? Sorry for not spotting the class something must have happened with my patch program. The problem is that I can't reproduce the reported behaviour with this test, all tests just run fine. The statements swap seems reasonable and I can definitely do it but I'd like to see a test fail first and pass after, any idea? Mar 31, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) commented on by goo...@anycpu.com - The original patch had the source of class ConstructorRecursionHazardClass at the end. This class just calls Type.forInstance() in its constructor. I'm pasting the pertinent contents of the patch file below. "Certain situatins" means that the recursion occurs whenever an inspected class in its constructor uses Type.forClass() directly or indirectly. +++ test/actionscript/as3reflect/testclasses/ConstructorRecursionHazardClass.as (revision 0) @@ -0,0 +1,37 @@ +/* + * Copyright (c) 2007-2009 the original author or authors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + / +package as3reflect.testclasses { + import as3reflect.Type; + + /* + * @author JÃ¼rgen Failenschmid + / + public class ConstructorRecursionHazardClass { + + public var varPublic:String = "varPublic"; + + public function ConstructorRecursionHazardClass(arg:String = null) { + // this could lead to infinite recursion - see http://code.google.com/p/as3reflect/issues/detail?id=5 + var type:Type = Type.forInstance(this); + } + } +} The original patch had the source of class ConstructorRecursionHazardClass at the end. This class just calls Type.forInstance() in its constructor. I'm pasting the pertinent contents of the patch file below. "Certain situatins" means that the recursion occurs whenever an inspected class in its constructor uses Type.forClass() directly or indirectly. +++ test/actionscript/as3reflect/testclasses/ConstructorRecursionHazardClass.as (revision 0) @@ -0,0 +1,37 @@ +/ + * Copyright (c) 2007-2009 the original author or authors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + / +package as3reflect.testclasses { + import as3reflect.Type; + + /* + * @author JÃ¼rgen Failenschmid + / + public class ConstructorRecursionHazardClass { + + public var varPublic:String = "varPublic"; + + public function ConstructorRecursionHazardClass(arg:String = null) { + // this could lead to infinite recursion - see http://code.google.com/p/as3reflect/issues/detail?id=5 + var type:Type = Type.forInstance(this); + } + } +} Mar 31, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) Status changed by martino.piccinato - Your patch is missing ConstructorRecursionHazard class. I tried to complete the test in the patch attached to this comment but could not reproduce the problem. Could you give some more context information. what you mean exactly with "certain situations"? Note that we moved the project to http://code.google.com/p/as3-commons/ so the patch is done against latest trunk of source code you see there (we are trying to close the last tickets here before releasing 1.0). Status: Started Your patch is missing ConstructorRecursionHazard class. I tried to complete the test in the patch attached to this comment but could not reproduce the problem. Could you give some more context information. what you mean exactly with "certain situations"? Note that we moved the project to http://code.google.com/p/as3-commons/ so the patch is done against latest trunk of source code you see there (we are trying to close the last tickets here before releasing 1.0). Status: Started Mar 31, 2009 issue 3 (AS3Reflect in maven repository) Status changed by martino.piccinato - AS3Reflect is now hosted at yoolab maven repository https://dev.yoolab.org/maven/index.html (releases and snapshots) More informations on http://www.as3commons.org/ as AS3Reflect is now a subproject of AS3Commons. Status: Fixed AS3Reflect is now hosted at yoolab maven repository https://dev.yoolab.org/maven/index.html (releases and snapshots) More informations on http://www.as3commons.org/ as AS3Reflect is now a subproject of AS3Commons. Status: Fixed Mar 15, 2009 issue 7 (Potential solution for classUtils::newInstance(clazz,args)) reported by Fordiman - I came to your as3reflect class in the hopes that you'd have a better solution for wrapped instantiation of new classes. Unfortunately, your code has the same switch mine does (cries). I came across Metal Hurlant's eval() work though, and it got me thinking: Obviously, an eval() in AS3 would involve an AS3 parser / compiler with AS3 bytecode as the target, then somehow run the AS3 bytecode. If there's a way to run arbitrary bytecode, then we could, theoretically, construct the bytecode* for creating a new instance and execute it, all within the newInstance function. So it then occurred to me that one could, theoretically, write the actual instantiation function in AS3 assembly, were that possible. Unfortunately, in this investigation, I found that the construct opcode must have a static byte telling it how many arguments to pull from the stack. As such, again, we absolutely HAVE to build the function at runtime, and execute it. I'm looking into how the ASC compiler in AS3 executes arbitrary bytecode. If I get it working, I intend to share how it's done. I came to your as3reflect class in the hopes that you'd have a better solution for wrapped instantiation of new classes. Unfortunately, your code has the same switch mine does (cries). I came across Metal Hurlant's eval() work though, and it got me thinking: Obviously, an eval() in AS3 would involve an AS3 parser / compiler with AS3 bytecode as the target, then somehow run the AS3 bytecode. If there's a way to run arbitrary bytecode, then we could, theoretically, construct the bytecode for creating a new instance and execute it, all within the newInstance function. So it then occurred to me that one could, theoretically, write the actual instantiation function in AS3 assembly, were that possible. Unfortunately, in this investigation, I found that the construct opcode must have a static byte telling it how many arguments to pull from the stack. As such, again, we absolutely HAVE to build the function at runtime, and execute it. I'm looking into how the ASC compiler in AS3 executes arbitrary bytecode. If I get it working, I intend to share how it's done. Mar 07, 2009 issue 6 (Type.forClass may enter infinite recursion due to describeTy...) reported by goo...@anycpu.com - See attached patch file for the trunk (as of 3-7-2009 8pm PST). What steps will reproduce the problem? 1. Run unit test TypeTest#testConstructorRecursionHazard. 2. If the queried class has a constructor that uses Type#forClass (directly or indirectly), a stack overflow occurs. This error is caught by a try...catch block in _getTypeDescription, therefore the issue is not observable externally, but in certain situations the stack overflow leads to partially or fully empty Type instances. The attached patch fixes the problem by swapping some statements in Type#forClass. See attached patch file for the trunk (as of 3-7-2009 8pm PST). What steps will reproduce the problem? 1. Run unit test TypeTest#testConstructorRecursionHazard. 2. If the queried class has a constructor that uses Type#forClass (directly or indirectly), a stack overflow occurs. This error is caught by a try...catch block in _getTypeDescription, therefore the issue is not observable externally, but in certain situations the stack overflow leads to partially or fully empty Type instances. The attached patch fixes the problem by swapping some statements in Type#forClass. Feb 20, 2009 issue 3 (AS3Reflect in maven repository) commented on by wangping...@163.com - is maven 2 ? very nice ! is maven 2 ? very nice ! Feb 20, 2009 r29 (Added package assembly to produce destribution zip. Added de...) committed by martino.piccinato - Added package assembly to produce destribution zip. Added deploy and release descriptors in pom. Added package assembly to produce destribution zip. Added deploy and release descriptors in pom. Feb 19, 2009 r28 (Reorganized directory structure to be maven compatible. Adde...) committed by martino.piccinato - Reorganized directory structure to be maven compatible. Added basic pom.xml. Reorganized directory structure to be maven compatible. Added basic pom.xml. Jan 26, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) Labels changed by christophe.herreman - We can remove the trace. However this will result in reference errors failing silently. Since these errors occur in Type.forName on inner class class declarations I think that's fine. Using the logging framework would result in a Flex framework dependency which is something we need to prevent with this project. I'm curious to know however if a logging API is available that is AS3 compatible. Labels: Type-Other Priority-Low −Type-Defect −Priority-Medium We can remove the trace. However this will result in reference errors failing silently. Since these errors occur in Type.forName on inner class class declarations I think that's fine. Using the logging framework would result in a Flex framework dependency which is something we need to prevent with this project. I'm curious to know however if a logging API is available that is AS3 compatible. Labels: Type-Other Priority-Low −Type-Defect −Priority-Medium Jan 26, 2009 issue 2 (accessor declaredBy type is matching the class type) changed by christophe.herreman - fixed and verified in TypeTest Status: Fixed Owner: christophe.herreman fixed and verified in TypeTest Status: Fixed Owner: christophe.herreman Jan 26, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) Status changed by christophe.herreman - Status: Accepted Status: Accepted Jan 26, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) changed by christophe.herreman - fixed and verified in TypeTest thx for reporting! Status: Fixed Owner: christophe.herreman fixed and verified in TypeTest thx for reporting! Status: Fixed Owner: christophe.herreman Jan 26, 2009 r27 (added PublicSubClass) committed by christophe.herreman - added PublicSubClass added PublicSubClass Jan 26, 2009 r26 (added testForClass_shouldHaveCorrectDeclaringTypesOnAccessor...) committed by christophe.herreman - added testForClass_shouldHaveCorrectDeclaringTypesOnAccessors test added testForClass_shouldHaveCorrectDeclaringTypesOnAccessors test Jan 26, 2009 r25 (fixed "parseAccessorsByModifier" in TypeParser so that decla...) committed by christophe.herreman - fixed "parseAccessorsByModifier" in TypeParser so that declaring type is correct fixed "parseAccessorsByModifier" in TypeParser so that declaring type is correct Jan 26, 2009 issue 4 (hasMetaData method in MetaDataContainer makes the wrong chec...) Status changed by christophe.herreman - fixed and verified in MetaDataContainerTest Status: Fixed fixed and verified in MetaDataContainerTest Status: Fixed Jan 26, 2009 r24 (added MetaDataContainerTest) committed by christophe.herreman - added MetaDataContainerTest added MetaDataContainerTest Jan 26, 2009 r23 (fixed "hasMetaData" method) committed by christophe.herreman - fixed "hasMetaData" method fixed "hasMetaData" method Jan 26, 2009 r22 (updated changelog) committed by christophe.herreman - updated changelog updated changelog Jan 26, 2009 r21 (updated changelog) committed by christophe.herreman - updated changelog updated changelog Jan 26, 2009 issue 5 (MethodInvoker does not work on proxied methods) Status changed by christophe.herreman - fixed and verified with MethodInvokerTest Status: Fixed fixed and verified with MethodInvokerTest Status: Fixed Jan 26, 2009 r20 (added MethodInvokerTest) committed by christophe.herreman - added MethodInvokerTest added MethodInvokerTest Jan 26, 2009 r19 (fixed import of ComplexClass) committed by christophe.herreman - fixed import of ComplexClass fixed import of ComplexClass Jan 26, 2009 r18 (removed assertion since this is a Spring ActionScript depend...) committed by christophe.herreman - removed assertion since this is a Spring ActionScript dependency removed assertion since this is a Spring ActionScript dependency Jan 26, 2009 r17 (added Proxy subclass for testing) committed by christophe.herreman - added Proxy subclass for testing added Proxy subclass for testing Jan 26, 2009 r16 ("invoke" method now uses MethodInvoker) committed by christophe.herreman - "invoke" method now uses MethodInvoker "invoke" method now uses MethodInvoker Jan 26, 2009 r15 (fixed "invoke" method so that proxied methods are invoked co...) committed by christophe.herreman - fixed "invoke" method so that proxied methods are invoked correctly fixed "invoke" method so that proxied methods are invoked correctly Jan 26, 2009 issue 5 (MethodInvoker does not work on proxied methods) reported by christophe.herreman - What steps will reproduce the problem? 1. create a MethodInvoker for invoking a method on a remote object 2. invoke the MethodInvoker What is the expected output? What do you see instead? - expected: normal execution of the remote call - actual: no method is found, because Type does not recognize proxied methods -> don't use type.getMethod() What steps will reproduce the problem? 1. create a MethodInvoker for invoking a method on a remote object 2. invoke the MethodInvoker What is the expected output? What do you see instead? - expected: normal execution of the remote call - actual: no method is found, because Type does not recognize proxied methods -> don't use type.getMethod() Jan 25, 2009 issue 4 (hasMetaData method in MetaDataContainer makes the wrong chec...) reported by s...@dds.nl - hasMetaData () currently has this implementation: public function hasMetaData(key:String):Boolean { return (getMetaData(key) != null); } this should be changed to: public function hasMetaData(key:String):Boolean { return (getMetaData(key).length > 0); } since getMetaData now always returns an array of metadata for the given key, and an empty array if the key doesn't exist. hasMetaData () currently has this implementation: public function hasMetaData(key:String):Boolean { return (getMetaData(key) != null); } this should be changed to: public function hasMetaData(key:String):Boolean { return (getMetaData(key).length > 0); } since getMetaData now always returns an array of metadata for the given key, and an empty array if the key doesn't exist. Jan 24, 2009 issue 3 (AS3Reflect in maven repository) reported by Sena.Pedro - I think that the title is enough, right? :) Would be very nice to have it in a maven repository. Regards, Pedro I think that the title is enough, right? :) Would be very nice to have it in a maven repository. Regards, Pedro Jan 23, 2009 issue 2 (accessor declaredBy type is matching the class type) reported by intonarumori - TypeXmlParser.parseAccessorsByModifier should create the Accessor instance with Type.forName(accessorXML.@declaredBy) instead of type in the declareType field. TypeXmlParser.parseAccessorsByModifier should create the Accessor instance with Type.forName(accessorXML.@declaredBy) instead of type in the declareType field. Jan 21, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) commented on by goo...@anycpu.com - Here's a log of starting one of our applications (that uses as3reflect via Spring ActionScript). It appears that all of these are caused by internal classes defined in the same source file as some public class. Type.forName error: Error #1065: Variable MapOfMaps is not defined. The class 'Logger.as$409::MapOfMaps' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable LoadDescription is not defined. The class 'Logger.as$409::LoadDescription' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable ListCollectionViewBookmark is not defined. The class 'ListCollectionView.as$73::ListCollectionViewBookmark' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable DataListRequestResponder is not defined. The class 'ConcreteDataService.as$79::DataListRequestResponder' is probably an internal class or it may not have been compiled. Here's a log of starting one of our applications (that uses as3reflect via Spring ActionScript). It appears that all of these are caused by internal classes defined in the same source file as some public class. Type.forName error: Error #1065: Variable MapOfMaps is not defined. The class 'Logger.as$409::MapOfMaps' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable LoadDescription is not defined. The class 'Logger.as$409::LoadDescription' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable ListCollectionViewBookmark is not defined. The class 'ListCollectionView.as$73::ListCollectionViewBookmark' is probably an internal class or it may not have been compiled. Type.forName error: Error #1065: Variable DataListRequestResponder is not defined. The class 'ConcreteDataService.as$79::DataListRequestResponder' is probably an internal class or it may not have been compiled. Jan 21, 2009 r14 (updated changelog) committed by christophe.herreman - updated changelog updated changelog Jan 21, 2009 r13 (added Constructor support) committed by christophe.herreman - added Constructor support added Constructor support Jan 20, 2009 issue 1 (Type.forName() contains a trace() call, that cannot be filte...) reported by goo...@anycpu.com - The application must be able to filter all log entries. What steps will reproduce the problem? 1. Call Type.forName() for an undefined class. 2. Observe an unconditional trace log entry. For example, this happens for classes generated by Flex: Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. This is not really an error, it should be treated as a warning. Instead of the trace() call, a logging framework should be used (level "Warn"), so that the application can decide if these kind of entries should be included in the log. What version of the product are you using? On what operating system? as3reflect as of 1-20-09 The application must be able to filter all log entries. What steps will reproduce the problem? 1. Call Type.forName() for an undefined class. 2. Observe an unconditional trace log entry. For example, this happens for classes generated by Flex: Type.forName error: Error #1065: Variable AuthenticationAgent is not defined. The class 'ChannelSet.as$125::AuthenticationAgent' is probably an internal class or it may not have been compiled. This is not really an error, it should be treated as a warning. Instead of the trace() call, a logging framework should be used (level "Warn"), so that the application can decide if these kind of entries should be included in the log. What version of the product are you using? On what operating system? as3reflect as of 1-20-09 Jan 20, 2009 as3reflect_1.0RC2.zip (as3reflect 1.0 RC2) file uploaded by christophe.herreman Jan 13, 2009 r12 (added changelog) committed by christophe.herreman - added changelog added changelog Jan 13, 2009 r11 (added testcases) committed by christophe.herreman - added testcases added testcases Jan 13, 2009 r10 (fixed indentations) committed by christophe.herreman - fixed indentations fixed indentations Jan 13, 2009 r9 (code style) committed by christophe.herreman - code style code style Jan 13, 2009 r8 (constructor now creates empty arguments array when null is p...) committed by christophe.herreman - constructor now creates empty arguments array when null is passed constructor now creates empty arguments array when null is passed Jan 13, 2009 r7 (IMetaDataContainer.getMetaData() now returns an array of Met...) committed by christophe.herreman - IMetaDataContainer.getMetaData() now returns an array of MetaData objects IMetaDataContainer.getMetaData() now returns an array of MetaData objects Jan 13, 2009 r6 (IMetaDataContainer.getMetaData() now returns an array of Met...) committed by christophe.herreman - IMetaDataContainer.getMetaData() now returns an array of MetaData objects IMetaDataContainer.getMetaData() now returns an array of MetaData objects Jan 10, 2009 as3reflect_1.0RC1.zip (as3reflect 1.0 RC1) file uploaded by christophe.herreman - Labels: OpSys-All Labels: OpSys-All

issue 6 (Type.forClass may enter infinite recursion due to describeTy...) commented on by goo...@anycpu.com - The original patch had the source of class ConstructorRecursionHazardClass at the end. This class just calls Type.forInstance() in its constructor. I'm pasting the pertinent contents of the patch file below. "Certain situatins" means that the recursion occurs whenever an inspected class in its constructor uses Type.forClass() directly or indirectly. +++ test/actionscript/as3reflect/testclasses/ConstructorRecursionHazardClass.as (revision 0) @@ -0,0 +1,37 @@ +/* + * Copyright (c) 2007-2009 the original author or authors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ +package as3reflect.testclasses { + import as3reflect.Type; + + /** + * @author JÃ¼rgen Failenschmid + */ + public class ConstructorRecursionHazardClass { + + public var varPublic:String = "varPublic"; + + public function ConstructorRecursionHazardClass(arg:String = null) { + // this could lead to infinite recursion - see http://code.google.com/p/as3reflect/issues/detail?id=5 + var type:Type = Type.forInstance(this); + } + } +}

The original patch had the source of class ConstructorRecursionHazardClass at the end. This class just calls Type.forInstance() in its constructor. I'm pasting the pertinent contents of the patch file below. "Certain situatins" means that the recursion occurs whenever an inspected class in its constructor uses Type.forClass() directly or indirectly. +++ test/actionscript/as3reflect/testclasses/ConstructorRecursionHazardClass.as (revision 0) @@ -0,0 +1,37 @@ +/* + * Copyright (c) 2007-2009 the original author or authors + * + * Permission is hereby granted, free of charge, to any person obtaining a copy + * of this software and associated documentation files (the "Software"), to deal + * in the Software without restriction, including without limitation the rights + * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the Software is + * furnished to do so, subject to the following conditions: + * + * The above copyright notice and this permission notice shall be included in + * all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN + * THE SOFTWARE. + */ +package as3reflect.testclasses { + import as3reflect.Type; + + /** + * @author JÃ¼rgen Failenschmid + */ + public class ConstructorRecursionHazardClass { + + public var varPublic:String = "varPublic"; + + public function ConstructorRecursionHazardClass(arg:String = null) { + // this could lead to infinite recursion - see http://code.google.com/p/as3reflect/issues/detail?id=5 + var type:Type = Type.forInstance(this); + } + } +}

Older