Fri, 14 Oct 2016 13:46:05 +0200 #REFACTORING by stefan cvs_MAIN
Stefan Vogel <sv@exept.de> [Fri, 14 Oct 2016 13:46:05 +0200] rev 3629
#REFACTORING by stefan class: JavaMethod added: #overwrites: removed: #overrides: be compatible with Method
Wed, 12 Oct 2016 23:02:07 +0200 #BUGFIX by mawalch cvs_MAIN
mawalch [Wed, 12 Oct 2016 23:02:07 +0200] rev 3628
#BUGFIX by mawalch class: JavaSettingsApplication changed: #javaHomeList Care for the case that the Java installation could not be located.
Tue, 20 Sep 2016 22:56:52 +0100 Issue #93 [3/3]: Avoid (expensive) registry scan to find all erroneous classes
Jan Vrany <jan.vrany@fit.cvut.cz> [Tue, 20 Sep 2016 22:56:52 +0100] rev 3627
Issue #93 [3/3]: Avoid (expensive) registry scan to find all erroneous classes Instead remember such classes in a separate set when the class is registered. The set is NOT weak and does not need to. Registry holds classes strongly anyway. There's no point to recompile an obsolete class (I believe, we'll see). This yields in another ~25% speedup for Groovy classloading benchmark. https://swing.fit.cvut.cz/projects/stx-libjava/ticket/93
Wed, 21 Sep 2016 09:18:20 +0100 Issue #93 [2/3]: Avoid (expensive) test for compilation error at (re)compilation time
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 21 Sep 2016 09:18:20 +0100] rev 3626
Issue #93 [2/3]: Avoid (expensive) test for compilation error at (re)compilation time Instead, detect compilation errors when .class file is read by `JavaClassReader`. If error are detected, mark class a erroneous using ACX_HASERRORS flag. At (re)compilation time, just test for this flag, which is a lot quicker than scanning a constant pool. https://swing.fit.cvut.cz/projects/stx-libjava/ticket/93
Tue, 20 Sep 2016 10:11:05 +0100 Issue #93 [1/3]: Added 2 simple benchmarks to measure performance of class loading
Jan Vrany <jan.vrany@fit.cvut.cz> [Tue, 20 Sep 2016 10:11:05 +0100] rev 3625
Issue #93 [1/3]: Added 2 simple benchmarks to measure performance of class loading https://swing.fit.cvut.cz/projects/stx-libjava/ticket/93
Mon, 19 Sep 2016 23:20:38 +0100 Update system classloader's classpath when a package is loaded
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 19 Sep 2016 23:20:38 +0100] rev 3624
Update system classloader's classpath when a package is loaded ...at runtime and defines its own java libraries. While this should not be a common case for standalone applications, in development mode once often load packages on demand. Then this comes handy as one does not need to reboot the JVM.
Mon, 19 Sep 2016 17:01:59 +0100 Fixed #isObsolete for Java classes being replaceb by newer version(s).
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 19 Sep 2016 17:01:59 +0100] rev 3623
Fixed #isObsolete for Java classes being replaceb by newer version(s). * Fixed JavaClass>>isObsolete to test for bit ACX_OBSOLETE. * Fixed JavaMetaclass>>isObsolete to ask it's class. * Changed `JavaClassRegistry` to mark old class as obsolete when it's replaced. This is necessary for tools (namely browser) as it refetches class the onecurrently being displayed is changed. This depends on correct return value of #isObsolete.
Mon, 19 Sep 2016 15:26:35 +0100 Renamed ACC_OBSOLETE to ACX_OBSOLETE as JVM spec does not define ACC_OBSOLETE
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 19 Sep 2016 15:26:35 +0100] rev 3622
Renamed ACC_OBSOLETE to ACX_OBSOLETE as JVM spec does not define ACC_OBSOLETE Define ACX_HASERRORS (not used currently)
Sun, 18 Sep 2016 12:11:12 +0100 Issue #49: Fixed a little bug when setting `thisContext` for debugger evaluation
Jan Vrany <jan.vrany@fit.cvut.cz> [Sun, 18 Sep 2016 12:11:12 +0100] rev 3621
Issue #49: Fixed a little bug when setting `thisContext` for debugger evaluation ...caused by refactoring. Test was missing so not spotted before. Sigh.
Tue, 27 Sep 2016 15:18:24 +0200 #DOCUMENTATION by mawalch cvs_MAIN
mawalch [Tue, 27 Sep 2016 15:18:24 +0200] rev 3620
#DOCUMENTATION by mawalch class: JavaMethod comment/format in: #resolve
(0) -3000 -1000 -300 -100 -10 +10 +100 +300 tip