rt/vm/src/main/java/org/apidesign/vm4brwsr/ByteCodeToJavaScript.java
Mon, 26 May 2014 14:33:11 +0200 Need to flush the stack before destroying it with return value of a function
Thu, 22 May 2014 15:29:40 +0200 Tests pass OK in full obfuscation mode
Tue, 13 May 2014 13:34:14 +0200 Use string names for annotation properties and exported methods to define enums
Sun, 11 May 2014 23:06:45 +0200 Full obfuscation of libraries works OK
Wed, 07 May 2014 11:41:41 +0200 Encode package-info into package_0002dinfo
Wed, 07 May 2014 10:01:54 +0200 Shortening constructor calls and access to arrays
Tue, 06 May 2014 11:12:40 +0200 Storing constructor function in last element of multidimensional array
Tue, 06 May 2014 10:08:42 +0200 2nd step in eliminating need for Class.forName when dealing with arrays
Tue, 06 May 2014 08:36:54 +0200 1st step in eliminating Class.forName when working with array types. Class.getComponentType on directly allocated arrays does not need that.
Sun, 04 May 2014 12:02:57 +0200 Exporting widely used methods via Class, which is guaranteed to be available, rather than Throwable
Thu, 01 May 2014 13:43:36 +0200 Don't reference Fn
Wed, 30 Apr 2014 15:57:56 +0200 Mangling class names, so UnderTest passes
Wed, 30 Apr 2014 15:04:10 +0200 Merging from default branch and resolving conflicts. mvn install -DskipTests passes OK.
Tue, 29 Apr 2014 16:28:06 +0200 Decreasing the number of lines of core bck2brwsr.js to be under 65636. Should help debugging on Nashorn as it has problems with lines numbers over 16-bit numbers.
Tue, 29 Apr 2014 16:22:23 +0200 Making ByteCodeToJavaScript an Appendable to track when something is generated
Thu, 24 Apr 2014 18:05:21 +0200 Backporting da9e5973e699 to make sure we run on Nashorn (at least a bit)
Tue, 18 Feb 2014 22:29:25 +0100 Shortening the generated code by providing shorter new array methods
Tue, 18 Feb 2014 22:20:50 +0100 More effective replaces on stack
Tue, 18 Feb 2014 10:02:46 +0100 Don't generate String constants either
Mon, 17 Feb 2014 17:41:00 +0100 No need to generate primitive final fields as their values get inlined in the classfiles anyway
Mon, 17 Feb 2014 15:39:34 +0100 15KB less from 1.4MB by eliminating empty for operations
Mon, 17 Feb 2014 14:26:55 +0100 Another if statements to flush stack before generating them
Mon, 17 Feb 2014 07:54:35 +0100 Don't clear the stack mapping when doing a cast
Sun, 16 Feb 2014 23:56:01 +0100 One less ;
Sun, 16 Feb 2014 23:33:57 +0100 Use normal constants and only wrap them before calling an operation on them
Sun, 16 Feb 2014 22:22:22 +0100 Making sure bit manipulation with longs produce reasonable results
Sat, 15 Feb 2014 23:16:11 +0100 Flush stack before performing goto
Sat, 15 Feb 2014 23:13:53 +0100 Enclose constants as objects. Get the function arguments without creates stack objects. Flush stack on frame closing.
Sat, 15 Feb 2014 20:18:26 +0100 Eliminates use of stack in empty super constructor call
Sat, 15 Feb 2014 17:23:24 +0100 Keep the state of registers in an array and flush it later
Sat, 15 Feb 2014 14:36:43 +0100 Centralizing the stack pushes into single assign method
Thu, 09 Jan 2014 21:19:44 +0100 JavaScriptBody callbacks now use simplified parameter types - all objects are just java.lang.Object
Thu, 09 Jan 2014 08:21:26 +0100 Copy the array before passing it to JavaScript when using net.java.html.js.JavaScriptBody
Thu, 07 Nov 2013 09:41:45 +0100 _ in class and method names should be manged to _1
Wed, 30 Oct 2013 17:15:23 +0100 Adopting to JDK8's Nashorn differences. Most tests should now pass with Nashorn now.
Wed, 09 Oct 2013 22:03:26 +0200 Prefix well-known static enum fields with fld_
Wed, 09 Oct 2013 16:45:28 +0200 Prefix also static fields to avoid their confusion with keywords
Thu, 05 Sep 2013 09:36:06 +0200 Need to make sure an object implements all transitive interfaces, not just those specified directly
Fri, 12 Jul 2013 16:06:38 +0200 Needs to mangle the parameters to convert / to _
Fri, 12 Jul 2013 11:07:58 +0200 Make sure callback with parameters works. Plus get ready to usage of $ sign in $JsCallbacks$ naming scheme.
Thu, 11 Jul 2013 22:36:01 +0200 Recognizes callback to static Java methods
Thu, 11 Jul 2013 20:33:19 +0200 net.java.html's JavaScriptBody annotation is handled with the help of $JsCallbacks$ generated helper class
Wed, 10 Jul 2013 13:31:00 +0200 net.java.html.js.JavaScriptBody may contain javacall attribute
Tue, 09 Jul 2013 21:06:35 +0200 Basic support for net.java.html.js annotations
Fri, 24 May 2013 18:04:55 +0200 Merge with trunk
Mon, 20 May 2013 10:59:47 +0200 Obfuscation fixes to allow run demo-calculator as an external module
Mon, 13 May 2013 18:54:50 +0200 Changed the way the external classes are identified
Tue, 07 May 2013 19:01:14 +0200 Access non-final exported methods through "invoker"
Mon, 06 May 2013 18:06:08 +0200 Overridable class member access type
Mon, 06 May 2013 11:57:29 +0200 Removed ObfuscationDelegate
Fri, 03 May 2013 18:15:47 +0200 Generate $instOf in exported form
Fri, 26 Apr 2013 18:48:34 +0200 Partially working extension modules
Fri, 26 Apr 2013 13:35:21 +0200 #4669 @ html.java.net: Provide generic conversion function and when catching something, use it to convert to appropriate Java exception
Thu, 25 Apr 2013 16:17:48 +0200 Initial structure for extension modules
Wed, 10 Apr 2013 10:08:18 +0200 Throw real ClassCastException
Mon, 25 Mar 2013 13:33:03 +0100 Merging Lubo's closure work into default branch. Release 0.5 will have (partial) support for effective static compilation
Mon, 25 Mar 2013 12:47:25 +0100 For debugging purposes it is much better to give class constructor functions fully qualified names. Obfuscation will optimize them away later anyway.
Thu, 21 Mar 2013 18:48:46 +0100 Initial attempt for advanced obfuscation
Wed, 20 Mar 2013 21:24:36 +0100 Jump instructions need to use signed and constant pool referencing instructions unsigned short. Also getting ready for reading signed short composed from (improperly) two unsigned bytes.
Tue, 19 Mar 2013 13:18:02 +0100 Merge with trunk