Wed, 30 Mar 2016 21:45:33 +0100 Tests: Do spawn a separate JVM when running tests using ant and real JVM
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 30 Mar 2016 21:45:33 +0100] rev 3545
Tests: Do spawn a separate JVM when running tests using ant and real JVM Otherwise, the classpath is not honoured, causing certain tests to fail.
Wed, 30 Mar 2016 18:53:20 +0100 Tests: Moved `stx.libjava.tests.ExceptionTests` classes to mocks package
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 30 Mar 2016 18:53:20 +0100] rev 3544
Tests: Moved `stx.libjava.tests.ExceptionTests` classes to mocks package ...i.e., to `stx:libjava.tests.mocks.JavaExceptionTestsHelper. This makes it easier for ant task `test` to locate actual test cases and exclude mocks and test helpers.
Sun, 27 Mar 2016 00:41:25 +0000 Use #allPackageIDs instead of #allProjectIDs
Jan Vrany <jan.vrany@fit.cvut.cz> [Sun, 27 Mar 2016 00:41:25 +0000] rev 3543
Use #allPackageIDs instead of #allProjectIDs ...as the latter become obsolete. (grafted from fa52c6a8c459b8c3d2e6e4e45a0830ef7a342bdb)
Wed, 30 Mar 2016 18:26:28 +0100 Tests: Moved `stx.libjava.tests.asm.*` classes to mocks package
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 30 Mar 2016 18:26:28 +0100] rev 3542
Tests: Moved `stx.libjava.tests.asm.*` classes to mocks package ...i.e., to `stx:libjava.tests.mocks.asm`. This makes it easier for ant task `test` to locate actual test cases.
Sun, 20 Mar 2016 13:19:32 +0000 Build files: added ant target `test` to build and run tests.
Jan Vrany <jan.vrany@fit.cvut.cz> [Sun, 20 Mar 2016 13:19:32 +0000] rev 3541
Build files: added ant target `test` to build and run tests. Use `ant test` in java directory to run tests under real JVM. By default test sources should be in directory `java/src-tests` and are compiled to `java/bin-tests`. If `java/src-tests` does not exist, `ant test` is noop. Indeed, individual packages may add more test targets by extending targets `test.compile.pre`/`test.compile.post` and `test.run.pre` and \`test.run.post`. When one or more tests fail, property `test.faulure` should be set to indicate failure.
Mon, 21 Mar 2016 09:34:33 +0000 Fixed JavaClass>>isTestSelector: for dynamic ProxyMethod
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 21 Mar 2016 09:34:33 +0000] rev 3540
Fixed JavaClass>>isTestSelector: for dynamic ProxyMethod Must test for isJavaMethod before accessing JavaMethod's annotations. It could happen that the method is ProxyMethod (therefore a normal Smalltalk method) which also understands #annotations but the returned structure is completely different. Therefore we have to check to avoid possible DNU.
Mon, 21 Mar 2016 09:21:48 +0000 Fixed baaad bug in INVOKE? handling when method does not exist
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 21 Mar 2016 09:21:48 +0000] rev 3539
Fixed baaad bug in INVOKE? handling when method does not exist ...due to incompatible change in .class file. When a method that does not exist is about to be invoked, then `java.lang.NoSuchMethodError` must be thrown. If the instruction is INVOKEINTERFACE, `java.lang.AbstractMethodError` should be thrown instead. The VM does not check nor throw these exceptions by itself. Instead it relies on Smalltalk #doesNotUnderstand: mechanism. There it checks whether the sending context is a a JavaContext and if so, throw Java error as specified in JVM spec. Interesting how far we went without noticing...
Sun, 20 Mar 2016 23:55:04 +0000 Added tests for INVOKE? instruction when invoked method is missing
Jan Vrany <jan.vrany@fit.cvut.cz> [Sun, 20 Mar 2016 23:55:04 +0000] rev 3538
Added tests for INVOKE? instruction when invoked method is missing Since this test needs a special fiddling around. class files, the code is in separate directory and compiles to separate directory.
Tue, 15 Mar 2016 15:46:32 +0000 Added JavaMethod>>hasEnabledBreakpoints, required by new stx:libtool
Jan Vrany <jan.vrany@fit.cvut.cz> [Tue, 15 Mar 2016 15:46:32 +0000] rev 3537
Added JavaMethod>>hasEnabledBreakpoints, required by new stx:libtool (grafted from 76cf21b2a44156ae29689b67011159ae5393bf4f)
Sat, 12 Mar 2016 00:11:19 +0000 JavaByteCodeProcessor: fixed fetchBytes4 use #signExtendedLongValue
Jan Vrany <jan.vrany@fit.cvut.cz> [Sat, 12 Mar 2016 00:11:19 +0000] rev 3536
JavaByteCodeProcessor: fixed fetchBytes4 use #signExtendedLongValue ...instead of asSigned32. This should fix JavaByteCodeInterpreterTests>>testLookupSwitch
(0) -3000 -1000 -300 -100 -10 +10 +100 +300 tip