Community
Participate
Working Groups
Umbrella bug for addressing all 1.9 test case issues.
First set of fixes released via: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=9155dfe5fa69fb2096196e5313e43eeabab82363 This addresses all the failures in org.eclipse.jdt.core.tests.eval.TestAll. What's left now fall largely in two buckets: 1. Failures due to additional warning about deprecated stuff in Java 9. 2. Our JRT file system not considering javax.* modules, which are required by som tests.
(In reply to Jay Arthanareeswaran from comment #1) > 1. Failures due to additional warning about deprecated stuff in Java 9. This is fixed via: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=aee2925515ff92b7466c70c1632476d226c7aba8
These are the four tests currently failing: test008 - (org.eclipse.jdt.core.tests.compiler.regression.SerialVersionUIDTests) test009 - (org.eclipse.jdt.core.tests.compiler.regression.SerialVersionUIDTests) test010 - (org.eclipse.jdt.core.tests.compiler.regression.SerialVersionUIDTests) testBug386356_1 - (org.eclipse.jdt.core.tests.compiler.regression.AnnotationTest) Totally there are 26 failures across all compliance levels. Besides, the following two tests have been disabled for now and should be investigated: CodeSnippetTest.testPackage() NegativeCodeSnippetTest.testChangePackage()
Looks like the SerialVersionUIDTests tests are failing because our effort in loading classes from non java.base modules through our extended class loader is failing. This class loader is redirecting all requests for system classes to findSystemClass(). But my guess is in classpath mode, non java.base modules are not available. Perhaps there's a way; will check with the Jigsaw experts. I did a small experiment by commenting out the overridden loadClass in VerifyClassLoader and this makes all mentioned tests pass. But obviously this is done for some reason, proven by other failing tests.
(In reply to Jay Arthanareeswaran from comment #2) > (In reply to Jay Arthanareeswaran from comment #1) > > 1. Failures due to additional warning about deprecated stuff in Java 9. > > This is fixed via: > > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/ > ?h=BETA_JAVA9&id=aee2925515ff92b7466c70c1632476d226c7aba8 Jay I see the failure for ConditionalExpressionTest @ 8
(In reply to Manoj Palat from comment #5) > Jay I see the failure for ConditionalExpressionTest @ 8 This no longer fails. However, there are a bunch of failures in the following suites: org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests org.eclipse.jdt.compiler.apt.tests.FilerTests org.eclipse.jdt.compiler.apt.tests.NegativeTests Causing being the same in all cases. the failing test is: junit.framework.TestCase.assertEquals("Only one compiler available", 1, compilerCounter); in BatchTestUtils line number 288.
(In reply to Jay Arthanareeswaran from comment #6) Looks like there is a change in behavior of java.util.ServiceLoader.load(Class<T>) in JRE. In JRE 8, we get only one instance of this after loading EclipseCompiler with a Class.forName(), but in JRE 9 we get both JavacTool and EclipseCompiler. The EclipseCompiler is retrieved even when I comment out the Class.forName(). I wonder if the test should worry about this at all.
The following annotation related tests are failing: testProcessorArgumentsWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests) testCompilerOneClassWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests) testTwoAnnotations(org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests) testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.FilerTests) testNegativeModel10WithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.NegativeTests) The failure are caused by this commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=0171d3fa20410a884ea6de07ec393bc179fb73bd The compiler is calling reset() on the annotation processing manager after each round (in the same session, though) of compilation, which results in loosing the loaded processors. Needs a closer look.
There were test failures in the last Y build due to the hard-coded file path separator in the testcase. http://download.eclipse.org/eclipse/downloads/drops4/Y20160804-1000/testresults/html/org.eclipse.jdt.core.tests.compiler_ep47Y-unit-lin64_linux.gtk.x86_64_8.0.html Fixed with this commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=4f43bd149065ac93c6dd889b2740091a2111c0b4
One more failing test was found with JRE 9 and is fixed with: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=4853bfa39b4537870df1c5f0503a1078575fae46 and the failing tests @1.8 mentioned with comment #9 are disable for now: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?h=BETA_JAVA9&id=4853bfa39b4537870df1c5f0503a1078575fae46 At 1.9, there these failures: testInheritedAnnosWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests) testProcessorArgumentsWithSystemCompiler(org.eclipse.jdt.compiler.apt.tests.BatchDispatchTests) testTypeMirrorWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) testTypeMirrorWithSystemCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) testGenericsWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) testPackageAnnotationsWithJavac(org.eclipse.jdt.compiler.apt.tests.Java8ElementsTests) And whole suite is unable to run with JRE 9: org.eclipse.jdt.compiler.tool.tests.AllTests
The following tests are disabled for now: AnnotationTest#testBug386356_1 SerialVersionUIDTests#test008 SerialVersionUIDTests#test009 SerialVersionUIDTests#test010 These don't fail when run alone, but fail when run as part of the suite.
note: RunAllJava9Tests (ref bug 506952) can be used run all Java 9 tests
Locally running tests.compiler.regression on 9-ea+164 I observe these regressions: GenericTypeTest at 1.5+ - test1151, test1153, test1155 -> reflection now prints inners with '$' instead of expected '.' NullAnnotationBatchCompilerTest at 1.5+ - test440477, test440687b -> unexpected problems - test490010NoEeaFile3 -> different problem / severity LambdaExpressionTest at 1.8+ - test039, test447119d, test449063, test449063a, test449063e -> various errors SerializeLambdaTest at 1.8+ - testbug479119, testbug479119a -> InaccessibleObjectException XLargeTest at 9 - test0016, test0017, test0018, test0019 -> IllegalAccessError GenericsRegressionTest_1_8 at 1.8+: - testBug494733_comment0 -> unexpected warning EnumTest at 9 - test434442 -> IllegalAccessError Total: 51 failures I don't see these in gerrit nor Y-builds (running on 1.8?). Anybody else seeing these?
Locally running AllJavaModelTests on 9-ea+164 I see: - 7 errors - 520 failures Is that expected? Reproduced by anybody else?
(In reply to Stephan Herrmann from comment #14) > Locally running AllJavaModelTests on 9-ea+164 I see: > > - 7 errors > - 520 failures > > Is that expected? Reproduced by anybody else? FWIW, total # of tests was 7989 (running off of https://git.eclipse.org/r/#/c/93798/14) In a recent Y-build I see 7949, not sure if the diff is fully attributed to that gerrit change. Of course some tests still seem to pass when the don't find the expected JRE. Further note, that gerrit doesn't even tell us about AllJavaModelTests, since compiler tests already fail the build.
I see 7976 tests - 5 failed , rest passed. with ea 169, at commit bd6803034b95b7e0dd8c0cbcd0aead0a5c726f65 on windows
(In reply to Manoj Palat from comment #16) > I see 7976 tests - 5 failed , rest passed. > with ea 169, at commit bd6803034b95b7e0dd8c0cbcd0aead0a5c726f65 on windows Alright, on HEAD of BETA_JAVA9 things look better indeed, same numbers here. The problem exists only in https://git.eclipse.org/r/#/c/93798/ I checked if my #14 broke anything, but already at #13 I get 7 errors 520 failures.
(In reply to comment #13) > Locally running tests.compiler.regression on 9-ea+164 I observe these > regressions: > > GenericTypeTest at 1.5+ > - test1151, test1153, test1155 > -> reflection now prints inners with '$' instead of expected '.' > > NullAnnotationBatchCompilerTest at 1.5+ > - test440477, test440687b > -> unexpected problems > - test490010NoEeaFile3 > -> different problem / severity > > LambdaExpressionTest at 1.8+ > - test039, test447119d, test449063, test449063a, test449063e > -> various errors > > SerializeLambdaTest at 1.8+ > - testbug479119, testbug479119a > -> InaccessibleObjectException > > XLargeTest at 9 > - test0016, test0017, test0018, test0019 > -> IllegalAccessError > > GenericsRegressionTest_1_8 at 1.8+: > - testBug494733_comment0 > -> unexpected warning > > EnumTest at 9 > - test434442 > -> IllegalAccessError > > > Total: 51 failures > > I don't see these in gerrit nor Y-builds (running on 1.8?). > Anybody else seeing these? Yes, I believe we have bugs for each of these failures (In reply to comment #17) > (In reply to Manoj Palat from comment #16) > > I see 7976 tests - 5 failed , rest passed. > > with ea 169, at commit bd6803034b95b7e0dd8c0cbcd0aead0a5c726f65 on windows > > Alright, on HEAD of BETA_JAVA9 things look better indeed, same numbers here. > > The problem exists only in https://git.eclipse.org/r/#/c/93798/ > I checked if my #14 broke anything, but already at #13 I get 7 errors 520 > failures. I'm on it, I'm sure it must be a result of my changes to JavaSearchNameEnvironment
(In reply to Sasikanth Bharadwaj from comment #18) > (In reply to comment #13) > > Locally running tests.compiler.regression on 9-ea+164 I observe these > > regressions: > > > > GenericTypeTest at 1.5+ > > - test1151, test1153, test1155 > > -> reflection now prints inners with '$' instead of expected '.' > > > > NullAnnotationBatchCompilerTest at 1.5+ > > - test440477, test440687b > > -> unexpected problems > > - test490010NoEeaFile3 > > -> different problem / severity > > > > LambdaExpressionTest at 1.8+ > > - test039, test447119d, test449063, test449063a, test449063e > > -> various errors > > > > SerializeLambdaTest at 1.8+ > > - testbug479119, testbug479119a > > -> InaccessibleObjectException > > > > XLargeTest at 9 > > - test0016, test0017, test0018, test0019 > > -> IllegalAccessError > > > > GenericsRegressionTest_1_8 at 1.8+: > > - testBug494733_comment0 > > -> unexpected warning > > > > EnumTest at 9 > > - test434442 > > -> IllegalAccessError > > > > > > Total: 51 failures > > > > I don't see these in gerrit nor Y-builds (running on 1.8?). > > Anybody else seeing these? > Yes, I believe we have bugs for each of these failures I have fixes for some of these. Could you point me to the individual bugs so I can put the fixes into the right buckets? Question remains: are any gerrit or releng builds actually running on 9?
(In reply to comment #19) > I have fixes for some of these. Could you point me to the individual bugs so I > can put the fixes into the right buckets? > bug 509123 for GTT and LET bug 509753 for XLargeTest and EnumTest bug 509754 for SLT > Question remains: are any gerrit or releng builds actually running on 9? no, I run RunAllJava9Tests locally using jdk9 and rely on gerrit for other regressions, but even that is not sufficient as you pointed out
(In reply to Sasikanth Bharadwaj from comment #20) > (In reply to comment #19) > > I have fixes for some of these. Could you point me to the individual bugs so I > > can put the fixes into the right buckets? > > > bug 509123 for GTT and LET Fix released, but one issue remains (see bug 509123 comment 3): - commit 85bb2bb3 (bug 480992) promises specialized error messages mentioning the involved modules, which, however, doesn't seem to be implemented? Should we just revert the test change or is anybody working on such improved error message?
(In reply to Stephan Herrmann from comment #21) > (In reply to Sasikanth Bharadwaj from comment #20) > > (In reply to comment #19) > > > I have fixes for some of these. Could you point me to the individual bugs so I > > > can put the fixes into the right buckets? > > > > > bug 509123 for GTT and LET > > Fix released, but one issue remains (see bug 509123 comment 3): > - commit 85bb2bb3 (bug 480992) promises specialized error messages > mentioning the involved modules, which, however, doesn't seem to be > implemented? Should we just revert the test change or is anybody working on > such improved error message? Can anybody comment on the previous expected error message: "X (in module: Unnamed Module) cannot be cast to I (in module: Unnamed Module)" When writing bug 517808 comment 37 I was thinking of extending this message: The method serve(org.impl.SomeImpl) in the type Serve is not applicable for the arguments (org.impl.SomeImpl) to The method serve(mod.one/org.impl.SomeImpl) in the type Serve is not applicable for the arguments (mod.two/org.impl.SomeImpl) See also, that runtime stack traces already use the module/pack.age.Type syntax. *If* ever we introduce such syntax in error messages, we probably want this only, if otherwise the names would be identical, right?
I again see failures in these when run on 9: (In reply to Jay Arthanareeswaran from comment #10) > At 1.9, there these failures: > > [...] > testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) > testPackageAnnotationsWithJavac(org.eclipse.jdt.compiler.apt.tests. > Java8ElementsTests) Is this WIP, or s.t. I broke recently?
(In reply to Stephan Herrmann from comment #23) > I again see failures in these when run on 9: > > (In reply to Jay Arthanareeswaran from comment #10) > > At 1.9, there these failures: > > > > [...] > > testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) > > testPackageAnnotationsWithJavac(org.eclipse.jdt.compiler.apt.tests. > > Java8ElementsTests) > > Is this WIP, or s.t. I broke recently? I wouldn't worry about it for the moment. I mean, you needn't worry :) I plan to tackle them with 517841.
(In reply to Jay Arthanareeswaran from comment #24) > I wouldn't worry about it for the moment. I mean, you needn't worry :) > > I plan to tackle them with 517841. I should have mentioned, I have been seeing them at least as far back as July 6th.
(In reply to Jay Arthanareeswaran from comment #24) > (In reply to Stephan Herrmann from comment #23) > > I again see failures in these when run on 9: > > > > (In reply to Jay Arthanareeswaran from comment #10) > > > At 1.9, there these failures: > > > > > > [...] > > > testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) > > > testPackageAnnotationsWithJavac(org.eclipse.jdt.compiler.apt.tests. > > > Java8ElementsTests) > > > > Is this WIP, or s.t. I broke recently? > > I wouldn't worry about it for the moment. I mean, you needn't worry :) So I don't :)
Running on 9 I see NPE in these tests: - org.eclipse.jdt.core.tests.eval.CodeSnippetTest.testFor89632() - org.eclipse.jdt.core.tests.eval.NegativeCodeSnippetTest.testUnusedImport() I'm looking into these.
New Gerrit change created: https://git.eclipse.org/r/103333
(In reply to Eclipse Genie from comment #28) > New Gerrit change created: https://git.eclipse.org/r/103333 Found the cause in the fix for bug 519723: Path("List.class").endsWith(".class") is not true in java.nio.file. interesting API :-/
Gerrit change https://git.eclipse.org/r/103333 was merged to [BETA_JAVA9]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=dc5d985163d449aae2c8deb52fce01044c90baa3
If seen failures in ModuleCompilationTests.testBug519922 when run locally via RunAllJava9Tests. I haven't yet seen the problem when running only ModuleCompilationTests. Difference is that in org.eclipse.SomeInterface we sometimes complain org cannot be resolved to a type where we expect org.eclipse cannot be resolved to a type I haven't investigated this, nor do I think this is severe, just wanted to let others know.
(In reply to Jay Arthanareeswaran from comment #24) > (In reply to Stephan Herrmann from comment #23) > > I again see failures in these when run on 9: > > > > (In reply to Jay Arthanareeswaran from comment #10) > > > At 1.9, there these failures: > > > > > > [...] > > > testElementWithEclipseCompiler(org.eclipse.jdt.compiler.apt.tests.ModelTests) > > > testPackageAnnotationsWithJavac(org.eclipse.jdt.compiler.apt.tests. > > > Java8ElementsTests) > > > > Is this WIP, or s.t. I broke recently? > > I wouldn't worry about it for the moment. I mean, you needn't worry :) > > I plan to tackle them with 517841. For once I saw Java8ElementsTests.testTypeAnnotations12Binary() on gerrit (using 1.8 of course): java.lang.NullPointerException at org.eclipse.jdt.internal.compiler.apt.model.ElementsImpl.addMembers(ElementsImpl.java:216) at org.eclipse.jdt.internal.compiler.apt.model.ElementsImpl.getAllMembers(ElementsImpl.java:163) at org.eclipse.jdt.compiler.apt.tests.processors.elements.Java8ElementProcessor.tTypeAnnotations12(Java8ElementProcessor.java:605) at org.eclipse.jdt.compiler.apt.tests.processors.elements.Java8ElementProcessor.testTypeAnnotations12Binary(Java8ElementProcessor.java:641) I couldn't reproduce this locally, though.
(In reply to comment #31) > If seen failures in > ModuleCompilationTests.testBug519922 > when run locally via RunAllJava9Tests. I haven't yet seen the problem when > running only ModuleCompilationTests. > > Difference is that in org.eclipse.SomeInterface we sometimes complain > org cannot be resolved to a type > where we expect > org.eclipse cannot be resolved to a type > > I haven't investigated this, nor do I think this is severe, just wanted to let > others know. I see an additional failure in CastTest that's probably related to the commit http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=fce70ab5b1cb13a2f70c1b9062312a09892f1a96 Test expectation may need to be updated for 9
New Gerrit change created: https://git.eclipse.org/r/104068
(In reply to Sasikanth Bharadwaj from comment #33) > I see an additional failure in CastTest that's probably related to the commit > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=fce70ab5b1cb13a2f70c1b9062312a09892f1a96 Sorry, my bad. Must have missed to run the entire test class on 9 (and hudson doesn't catch this either). (In reply to Eclipse Genie from comment #34) > New Gerrit change created: https://git.eclipse.org/r/104068 > Test expectation may need to be updated for 9 The gerrit does exactly that.
Gerrit change https://git.eclipse.org/r/104068 was merged to [BETA_JAVA9]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=c3224ff6ed6b781979bda5c321c8dba2bc711589
Just to sync, this is the list of failures I get when running all JDT/Core tests on JRE 9 (total 82600): org.eclipse.jdt.core.tests.compiler.regression.XLargeTest.test0016 org.eclipse.jdt.core.tests.compiler.regression.XLargeTest.test0017 org.eclipse.jdt.core.tests.compiler.regression.XLargeTest.test0018 org.eclipse.jdt.core.tests.compiler.regression.XLargeTest.test0019 org.eclipse.jdt.core.tests.compiler.regression.EnumTest.test434442 org.eclipse.jdt.core.tests.builder.GetResourcesTests.test005 IIRC this is the best result I ever achieved in BETA_JAVA9 (5x IllegalAccessError in enum, 1x binary names for anonymous types).
I see an additional failure in ModuleAttributeTests.testModuleCompile when running RunAllJava9Tests - appears to be a line separator issue
(In reply to comment #38) > I see an additional failure in ModuleAttributeTests.testModuleCompile when > running RunAllJava9Tests - appears to be a line separator issue I meant path separator issue
(In reply to Sasikanth Bharadwaj from comment #39) > (In reply to comment #38) > > I see an additional failure in ModuleAttributeTests.testModuleCompile when > > running RunAllJava9Tests - appears to be a line separator issue > I meant path separator issue Sorry, my bad. As I can't test a fix (it's not broken on my machine), could you please try to fix this inside AbstractRegressionTest9.getCompilationUnits() by s.t. like fileNameString = fileNameString.replace(File.pathSeparator, '/') ? Looks easier, than having to concat with File.pathSeparator in each and every test that uses associateToModule(), OK? thanks
New Gerrit change created: https://git.eclipse.org/r/104682
(In reply to Eclipse Genie from comment #41) > New Gerrit change created: https://git.eclipse.org/r/104682 Fixes the failure in CompletionTests9.test486988_0015() A call to nameEnvironment.findType() was not module-aware, nor did we have API for clients outside the compiler to actually follow reads edges through the module graph. Added such function to SearchableEnvironment. Heads-up: CompletionTests9 is not yet listed below RunJDTCoreTests, only in RunAllJava9Tests! I do see Java9ElementsTest and ResolveTests9 in AllJavaModelTests, but this list isn't complete. Also missing: JavaSearchBugs9Tests. Anybody else? compiler.regression suite looks "more complete".
Gerrit change https://git.eclipse.org/r/104682 was merged to [BETA_JAVA9]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=44c3522e5259e25f9b607e9ab760cd16382d194e
Is anybody else seeing failure in ResolveTests.testInterfaceX()? I do, intermittently. Trouble shooting shows that the search gets cancelled since indexer is not ready. The call in question is searchAllTypeNames in SearchableEnvironment line number 572. I can force the tests to pass if I change the argument from CANCEL_IF_NOT_READY_TO_SEARCH to WAIT_UNTIL_READY_TO_SEARCH. Manoj, any idea what we can do here?
(In reply to Jay Arthanareeswaran from comment #44) > Is anybody else seeing failure in ResolveTests.testInterfaceX()? I do, > intermittently. Trouble shooting shows that the search gets cancelled since > indexer is not ready. The call in question is searchAllTypeNames in > SearchableEnvironment line number 572. > > I can force the tests to pass if I change the argument from > CANCEL_IF_NOT_READY_TO_SEARCH to WAIT_UNTIL_READY_TO_SEARCH. > > Manoj, any idea what we can do here? @Jay: On backtracking I found that this issue started coming up with commit 44c3522e5259e25f9b607e9ab760cd16382d194e. So you may want to start from these changes.
(In reply to comment #40) > (In reply to Sasikanth Bharadwaj from comment #39) > > (In reply to comment #38) > > > I see an additional failure in ModuleAttributeTests.testModuleCompile when > > > running RunAllJava9Tests - appears to be a line separator issue > > I meant path separator issue > > Sorry, my bad. > > As I can't test a fix (it's not broken on my machine), could you please try to > fix this inside AbstractRegressionTest9.getCompilationUnits() by s.t. like > fileNameString = fileNameString.replace(File.pathSeparator, '/') > ? > Looks easier, than having to concat with File.pathSeparator in each and every > test that uses associateToModule(), OK? > The quick fix didn't solve the issue, I will dig further and make the necessary changes when I add more tests to ModuleAttributeTests > thanks
(In reply to comment #46) > (In reply to comment #40) > > (In reply to Sasikanth Bharadwaj from comment #39) > > > (In reply to comment #38) > > > > I see an additional failure in ModuleAttributeTests.testModuleCompile when > > > > running RunAllJava9Tests - appears to be a line separator issue > > > I meant path separator issue > > > > Sorry, my bad. > > > > As I can't test a fix (it's not broken on my machine), could you please try to > > fix this inside AbstractRegressionTest9.getCompilationUnits() by s.t. like > > fileNameString = fileNameString.replace(File.pathSeparator, '/') > > ? > > Looks easier, than having to concat with File.pathSeparator in each and every > > test that uses associateToModule(), OK? > > > The quick fix didn't solve the issue, I will dig further and make the necessary > changes when I add more tests to ModuleAttributeTests > > thanks done via commit 26bb485688891a8824802195033c124ad5c82755. for bug 495967
Running builder tests with -Dcompliance=9 doesn't work. I get 21 errors from TestingEnvironment.addProject() like: This test requires a 1.5 JRE Obviously, a branch if ("9".equals(compliance)) is missing.
Anybody else seeing the failure in org.eclipse.jdt.core.tests.builder.GetResourcesTests.test005 ?
New Gerrit change created: https://git.eclipse.org/r/111029
(In reply to Stephan Herrmann from comment #48) > Running builder tests with -Dcompliance=9 doesn't work. > > I get 21 errors from TestingEnvironment.addProject() like: > This test requires a 1.5 JRE > > Obviously, a branch > if ("9".equals(compliance)) > is missing. that branch is missing, but that's not the cause for the exception. We seem to have different uses of -Dcompliance here. - Normally, in compiler tests this is used to select at which compliance the compiler should be tested. - In TestingEnvironment.addProject() the same property is, however, used to detect, whether a project can be created at a requested compliance level, and instead of skipping tests with unmet requirements, we throw RuntimeException. (In reply to Eclipse Genie from comment #50) > New Gerrit change created: https://git.eclipse.org/r/111029 this is the best & easiest fix I can think of right now: - add the missing branch - use java.specification.version rather than compliance - check "<" instead of flag inclusion. (Background: while gerrit jobs still run on Java 1.8 I need a way to locally perform all that is omitted, without waiting for results at all compliance levels).
Gerrit change https://git.eclipse.org/r/111029 was merged to [master]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=922e76f860ca2f7ff1a9623fc6eb87b3b2728f15
(In reply to Eclipse Genie from comment #52) > Gerrit change https://git.eclipse.org/r/111029 was merged to [master]. > Commit: > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=922e76f860ca2f7ff1a9623fc6eb87b3b2728f15 Just as I was going to write a final comment about this I realize that now I'm comparing apples (like ClassFileConstants.JDK9) with oranges (like AbstractCompilerTest.F_9) Additional fix coming soon ...
Created attachment 271457 [details] tracing output Dumping an interesting section from a hudson console. I particularly like this line: 06:58:16 !!! ERROR: P/P was never deleted even after having waited 0ms!!! Despite all that logging, the job seems to proceed alright, all this took roughly 2:30 minutes. Do we have an open bug that waits for this tracing output?
Updated list of problems (master): ModuleBuilderTests.testAutoModule4() - intermittent. I'll try to cover that via bug 522078 The 5 well-known IllegalAccessException re enum initialization ModuleCompilationTests.testBug519922() - random variation in error position, sometimes "org", sometimes "org.eclipse" GenericsRegressionTest_9.testBug488663_005() - random order in error message: Y<I & J> vs. Y<J & I>
Locally running RunJDTCoreTests (to workaround problems on hundson), I see these failures: - BuildPathTests.testIncompatibleJdkLEvelOnProject() - BuildPathTests.testIncompatibleJdkLEvelOnWksp() - GetResourcesTests.test005() plus 221 failures in various ASTConverter tests. Individually re-running these tests did not exhibit any failures, including RunConverterTests in its entirety. Anybody else seeing s.t. like this? All this is about running on 1.8.
(In reply to comment #55) > Updated list of problems (master): > > ModuleBuilderTests.testAutoModule4() > - intermittent. I'll try to cover that via bug 522078 > > The 5 well-known IllegalAccessException re enum initialization > > ModuleCompilationTests.testBug519922() > - random variation in error position, sometimes "org", sometimes "org.eclipse" > > GenericsRegressionTest_9.testBug488663_005() > - random order in error message: Y<I & J> vs. Y<J & I> I'm seeing a new failure - CompletionTests9.testBug522164_jar, failure trace as follows. Fails consistently whether run via RunAllJava9Tests or individually junit.framework.ComparisonFailure: ----------- Expected ------------ test[METHOD_REF]{test(), Lp.a.Ifc;, ()V, test, 60} ------------ but was ------------ --------- Difference is ---------- expected:<[test[METHOD_REF]{test(), Lp.a.Ifc;, ()V, test, 60}]> but was:<[]>
(In reply to Sasikanth Bharadwaj from comment #57) > I'm seeing a new failure - CompletionTests9.testBug522164_jar, failure trace > as follows. Fails consistently whether run via RunAllJava9Tests or > individually Test is failing from commit 2c4e46a161c87faff7d830d04cd9ce262b2a9e54 onwards
(In reply to Manoj Palat from comment #58) > (In reply to Sasikanth Bharadwaj from comment #57) > > > I'm seeing a new failure - CompletionTests9.testBug522164_jar, failure trace > > as follows. Fails consistently whether run via RunAllJava9Tests or > > individually > > Test is failing from commit 2c4e46a161c87faff7d830d04cd9ce262b2a9e54 onwards @Till: Would you be able to take a look on this?
(In reply to Manoj Palat from comment #59) > (In reply to Manoj Palat from comment #58) > > (In reply to Sasikanth Bharadwaj from comment #57) > > > > > I'm seeing a new failure - CompletionTests9.testBug522164_jar, failure trace > > > as follows. Fails consistently whether run via RunAllJava9Tests or > > > individually > > > > Test is failing from commit 2c4e46a161c87faff7d830d04cd9ce262b2a9e54 onwards > > @Till: Would you be able to take a look on this? Fix is on the way, I'm adding it to bug 527576.
(In reply to Till Brychcy from comment #60) > Fix is on the way, I'm adding it to bug 527576. Thanks Till for the fast response!
(In reply to Manoj Palat from comment #61) > (In reply to Till Brychcy from comment #60) > > > Fix is on the way, I'm adding it to bug 527576. > > Thanks Till for the fast response! Fix by Till: 2d19e28eb87a29fea3b4dab9b0289a073ee6f30c moving this out of M4 to address rest of test issues
New Gerrit change created: https://git.eclipse.org/r/113120
(In reply to Eclipse Genie from comment #63) > New Gerrit change created: https://git.eclipse.org/r/113120 Combines two fixes: (1) Enable running jenkins tests on Java 9 (see also bug 527750) - only applied to o.e.j.c.tests.compiler atm. (2) Avoid random order of interfaces in an intersection type that would cause random failures in GRT_9.testBug488663_005() (We were already sorting in order to put classes to front, the fix just piggy-backs on that).
(In reply to Stephan Herrmann from comment #64) > (2) Avoid random order of interfaces in an intersection type that would > cause random failures in GRT_9.testBug488663_005() > (We were already sorting in order to put classes to front, the fix just > piggy-backs on that). Including interfaces in the existing sort() in LE.createIntersectionType18() unvealed several code locations / tests that depended on existing order, bogusly I think. Since some of these changes are not obvious, here some notes on patch set #2: - Adjusted test expectation, LMK if you see problems with changed messages - Expression.computeConversion() accepted the well-known type Serializable as a legal boxing type leading to an illegal primitive cast instruction. - ReferenceExpression.generateCode() used the wrong type (not the FI) as the return type in invokedynamic A quick review would be great (https://git.eclipse.org/r/#/c/113120/)
Gerrit change https://git.eclipse.org/r/113120 was merged to [master]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=97508c79ae02c520c9c61b641f6ebc42e364ced7
(In reply to Eclipse Genie from comment #66) > Gerrit change https://git.eclipse.org/r/113120 was merged to [master]. > Commit: > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=97508c79ae02c520c9c61b641f6ebc42e364ced7 > Only after merging did I notice that the jenkins still used JRE 8. What am I missing?
(In reply to Jay Arthanareeswaran from comment #67) > (In reply to Eclipse Genie from comment #66) > > Gerrit change https://git.eclipse.org/r/113120 was merged to [master]. > > Commit: > > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=97508c79ae02c520c9c61b641f6ebc42e364ced7 > > > > Only after merging did I notice that the jenkins still used JRE 8. What am I > missing? The trick is: the build itself runs on JRE 8 (otherwise we would run into bug 514119. But the test execution uses JDK 9, see these log snippets: [INFO] ------------------------------------------------------------------------ [INFO] Building org.eclipse.jdt.core.tests.compiler 3.12.400-SNAPSHOT [INFO] ------------------------------------------------------------------------ ... [INFO] --- maven-toolchains-plugin:1.1:toolchain (default) @ org.eclipse.jdt.core.tests.compiler --- ... [INFO] Required toolchain: jdk [ id='JavaSE-9' ] [INFO] Found matching toolchain for type jdk: JDK[/opt/public/common/java/oracle/jdk-9-ga_x64] ... [INFO] --- tycho-surefire-plugin:1.1.0-SNAPSHOT:test (default-test) @ org.eclipse.jdt.core.tests.compiler --- [INFO] Toolchain in tycho-surefire-plugin: JDK[/opt/public/common/java/oracle/jdk-9-ga_x64] [INFO] Expected eclipse log file: /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/org.eclipse.jdt.core.tests.compiler/target/work/data/.metadata/.log [INFO] Command line: [/opt/public/common/java/oracle/jdk-9-ga_x64/bin/java, -Dosgi.noShutdown=false, -Dosgi.os=linux, -Dosgi.ws=gtk, -Dosgi.arch=x86_64, -Dosgi.clean=true, -jar, /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/.repository/p2/osgi/bundle/org.eclipse.equinox.launcher/1.4.0.v20161219-1356/org.eclipse.equinox.launcher-1.4.0.v20161219-1356.jar, -data, /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/org.eclipse.jdt.core.tests.compiler/target/work/data, -install, /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/org.eclipse.jdt.core.tests.compiler/target/work, -configuration, /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/org.eclipse.jdt.core.tests.compiler/target/work/configuration, -application, org.eclipse.tycho.surefire.osgibooter.headlesstest, -testproperties, /jobs/genie.jdt/eclipse.jdt.core-Gerrit/workspace/org.eclipse.jdt.core.tests.compiler/target/surefire.properties] ... Also see the increase in test numbers. Cool, huh? :)
(In reply to Stephan Herrmann from comment #68) > Also see the increase in test numbers. > > Cool, huh? :) Very cool, indeed!
Meanwhile the first jenkins run with model tests on JDK 9 [1] shows these failures: org.eclipse.jdt.core.tests.model.JavaProjectTests.testBug522554_2 org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testConvertToModule org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug512053 org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282 org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282a The first was expected, that's what drove me down the alley of testing on JDK 9 in the first place. Anyone recognize any of the failures in ModuleBuilderTests? [1] https://ci.eclipse.org/jdt/job/eclipse.jdt.core-Gerrit/158/testReport/
(In reply to Stephan Herrmann from comment #70) > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testConvertToModule > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug512053 > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282 > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282a They all seem to be about not being able to find the system classes. But need to dig deeper to find the cause. Also interesting to note that we are getting the options, but don't seem to be using them anywhere.
(In reply to Jay Arthanareeswaran from comment #71) > (In reply to Stephan Herrmann from comment #70) > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testConvertToModule > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug512053 > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282 > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282a > > They all seem to be about not being able to find the system classes. But > need to dig deeper to find the cause. > > Also interesting to note that we are getting the options, but don't seem to > be using them anywhere. I could not reproduce by invoking tests from Eclipse. So for good measure I retriggered the jenkins job (to check for cosmic interference) but the result was the same.
(In reply to Stephan Herrmann from comment #72) > (In reply to Jay Arthanareeswaran from comment #71) > > (In reply to Stephan Herrmann from comment #70) > > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testConvertToModule > > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug512053 > > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282 > > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testBug518282a > > > > They all seem to be about not being able to find the system classes. But > > need to dig deeper to find the cause. > > > > Also interesting to note that we are getting the options, but don't seem to > > be using them anywhere. > > I could not reproduce by invoking tests from Eclipse. So for good measure I > retriggered the jenkins job (to check for cosmic interference) but the > result was the same. Phew, this one is tricky to debug, here's an approach that works: We need to debug the test execution as it is invoked by maven/tycho, but due to the toolchains magic, tycho is forced to fork a new jvm for tests, hence running mvnDebug doesn't help. Instead we need to debug the target jvm launched by tycho: In tests.model/pom.xml extend the command line like so: <argLine>-Djdt.default.test.compliance=1.8 -agentlib:jdwp=transport=dt_socket,server=y,address=8000,suspend=y</argLine> Next, uncomment any test suites other than the one you want to debug. Run maven inside git/eclipse.jdt.core: $ mvn -o -P build-individual-bundles,bree-libs,test-on-javase-9 -pl org.eclipse.jdt.core.tests.model verify (-o to avoid waiting for "downloading the internet" ;p - requires a previous non-offline run) At some point the build will pause, saying Listening for transport dt_socket at address: 8000 Launch a remote debugging session in Eclipse (after having set relevant breakpoints of course). => Happy debugging.
(In reply to Stephan Herrmann from comment #73) > (In reply to Stephan Herrmann from comment #72) > > (In reply to Jay Arthanareeswaran from comment #71) > > > (In reply to Stephan Herrmann from comment #70) > > > > org.eclipse.jdt.core.tests.model.ModuleBuilderTests.testConvertToModule When debugging this one under maven etc. we are failing because in JavaModelManager.initializeContainer(IJavaProject, IPath) we don't find an initializer in path "org.eclipse.jdt.launching.JRE_CONTAINER/org.eclipse.jdt.internal.debug.ui.launcher.StandardVMType/JavaSE-9" Apparently, o.e.j.debug.lauching is not part of the SUT? Right, to reproduce in Eclipse do: - create launch config for ModuleBuilderTests - under Plug-ins: - deselect all - select o.e.j.core.tests.model - click "Add Required Plug-ins" - see that o.e.j.launching is not selected - run / debug Watch the same failures as reported above. => All those tests are bogus (when placed in o.e.j.core.tests.model), or we need to add a dependency on o.e.j.launching. Would the latter be legal?
(In reply to Stephan Herrmann from comment #74) > => All those tests are bogus (when placed in o.e.j.core.tests.model), or we > need to add a dependency on o.e.j.launching. Would the latter be legal? https://git.eclipse.org/r/#/c/113116/4 attempts to fix that problem by indeed just adding this one dependency to o.e.j.core.tests.model. (That change additionally contains improvements re bug 522554, so it's booked on that bug again).
(In reply to Stephan Herrmann from comment #74) > => All those tests are bogus (when placed in o.e.j.core.tests.model), or we > need to add a dependency on o.e.j.launching. Would the latter be legal? I was going to type out "No" then saw your next comment. So, will just wait for the tests. On the other hand, it would be okay since it is only a test project, right?
(In reply to Jay Arthanareeswaran from comment #76) > On the other hand, it would be okay since it is only a test > project, right? That's my hope. Otherwise we'd need some clever new design for our tests, or move some tests over to JDT/Debug. Let's see, but local maven build was happy with this change :) BTW: what's the other bug where we would have needed the JRE container, and therefor couldn't create the test in core-land? Or was that one not about compilation but about execution?
(In reply to Stephan Herrmann from comment #77) > BTW: what's the other bug where we would have needed the JRE container, and > therefor couldn't create the test in core-land? Or was that one not about > compilation but about execution? Bug 521995, comment #11 and #12.
(In reply to Jay Arthanareeswaran from comment #78) > (In reply to Stephan Herrmann from comment #77) > > BTW: what's the other bug where we would have needed the JRE container, and > > therefor couldn't create the test in core-land? Or was that one not about > > compilation but about execution? > > Bug 521995, comment #11 and #12. Thanks, your memory seems to work better than mine :) So, would the pending addition of a new dependency also unblock you in that bug?
(In reply to Stephan Herrmann from comment #79) > So, would the pending addition of a new dependency also unblock you in that > bug? I think so. I had earlier requested Sarika to help me in writing a test for that in debug, but if we can write such tests in our own test projects, then nothing like it.
(In reply to Stephan Herrmann from comment #68) > The trick is: the build itself runs on JRE 8 (otherwise we would run into > bug 514119. But the test execution uses JDK 9, see these log snippets: Stephan, one of the tests is failing on the patch for bug 490103. This isn't failing locally. I put some logs and this is what I found: Found matching toolchain for type jdk: JDK[/opt/public/common/java/oracle/jdk-9_x64-latest] Getting JRT from : /opt/public/common/java/oracle/jdk-9-ga_x64 As you can see, there are two JDKs, the second one retrieved from System.getProperty("java.home") I can't tell if the second is a proper JDK installation or not. But is there a way to get hold of the first JDK?
(In reply to Jay Arthanareeswaran from comment #81) > (In reply to Stephan Herrmann from comment #68) > > The trick is: the build itself runs on JRE 8 (otherwise we would run into > > bug 514119. But the test execution uses JDK 9, see these log snippets: > > Stephan, one of the tests is failing on the patch for bug 490103. This isn't > failing locally. I put some logs and this is what I found: > > Found matching toolchain for type jdk: > JDK[/opt/public/common/java/oracle/jdk-9_x64-latest] > Getting JRT from : /opt/public/common/java/oracle/jdk-9-ga_x64 > > As you can see, there are two JDKs, the second one retrieved from > System.getProperty("java.home") > > I can't tell if the second is a proper JDK installation or not. But is there > a way to get hold of the first JDK? As a first quick guess I would say that one path is a symlink to the other ("-latest" being the constant name the will get redirected when new versions are installed). FYI: if you need to inspect how things look in the filesystem on jenkins, adding a shell script as an additional build step on jenkins may come in handy :)
*** Bug 530212 has been marked as a duplicate of this bug. ***
(In reply to Stephan Herrmann from comment #83) > *** Bug 530212 has been marked as a duplicate of this bug. *** Hello! Not trying to be a pain, but could you please explain how this bug is a duplicate? I don't see the error I was getting anywhere in the comments. Is my scenario covered in one of the unit tests that have failed? Thanks!
(In reply to Ben Ferguson from comment #84) > (In reply to Stephan Herrmann from comment #83) > > *** Bug 530212 has been marked as a duplicate of this bug. *** > > Hello! Not trying to be a pain, but could you please explain how this bug is > a duplicate? I don't see the error I was getting anywhere in the comments. > Is my scenario covered in one of the unit tests that have failed? Thanks! I was surprised myself :) Did you see bug 530212 comment 1 and bug 530212 comment 2?
(In reply to Stephan Herrmann from comment #85) > (In reply to Ben Ferguson from comment #84) > > (In reply to Stephan Herrmann from comment #83) > > > *** Bug 530212 has been marked as a duplicate of this bug. *** > > > > Hello! Not trying to be a pain, but could you please explain how this bug is > > a duplicate? I don't see the error I was getting anywhere in the comments. > > Is my scenario covered in one of the unit tests that have failed? Thanks! > > I was surprised myself :) > Did you see bug 530212 comment 1 and bug 530212 comment 2? Ah, interesting. Gotta love it when issues accidentally fix themselves :). If I understand this correctly, this issue will be fixed in Oxygen v4.8?
(In reply to Ben Ferguson from comment #86) > (In reply to Stephan Herrmann from comment #85) > > (In reply to Ben Ferguson from comment #84) > > > (In reply to Stephan Herrmann from comment #83) > > > > *** Bug 530212 has been marked as a duplicate of this bug. *** > > > > > > Hello! Not trying to be a pain, but could you please explain how this bug is > > > a duplicate? I don't see the error I was getting anywhere in the comments. > > > Is my scenario covered in one of the unit tests that have failed? Thanks! > > > > I was surprised myself :) > > Did you see bug 530212 comment 1 and bug 530212 comment 2? > > Ah, interesting. Gotta love it when issues accidentally fix themselves :). > If I understand this correctly, this issue will be fixed in Oxygen v4.8? Yes, "your" bug is closed with target 4.8 M5.
New Gerrit change created: https://git.eclipse.org/r/116680
Gerrit change https://git.eclipse.org/r/116680 was merged to [master]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=dc9331a753b84c4de709d71cc785ab1fe2d10548
(In reply to Eclipse Genie from comment #89) > Gerrit change https://git.eclipse.org/r/116680 was merged to [master]. > Commit: > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/ > ?id=dc9331a753b84c4de709d71cc785ab1fe2d10548 This should fix the issue of JavaProjectTests.testBug522554_2 and JavaProjectTests.testBug522554_2_unlimited where we had trouble with knowing the set of default modules. Previously I thought this to be a matter of JDK version, but the more likely explanation is that Oracle JDK and OpenJDK simply have different sets.
(In reply to Stephan Herrmann from comment #90) > (In reply to Eclipse Genie from comment #89) > > Gerrit change https://git.eclipse.org/r/116680 was merged to [master]. > > Commit: > > http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/ > > ?id=dc9331a753b84c4de709d71cc785ab1fe2d10548 > > This should fix the issue of JavaProjectTests.testBug522554_2 and > JavaProjectTests.testBug522554_2_unlimited where we had trouble with knowing > the set of default modules. Previously I thought this to be a matter of JDK > version, but the more likely explanation is that Oracle JDK and OpenJDK > simply have different sets. Only now I found bug 529981, which I had raised specifically for this issue. Let's wait for the next I-build and if successful, close the other bug as a duplicate of this one.
*** Bug 529981 has been marked as a duplicate of this bug. ***
Finally, we have the missing test suites running with JRE 9: http://download.eclipse.org/eclipse/downloads/drops4/I20180207-2000/testresults/html/org.eclipse.jdt.compiler.apt.tests_ep48I-unit-cen64-gtk3-java9_linux.gtk.x86_64_9.html I see five failures in APT, which I can reproduce locally as well: Java9ElementsTests#testRootElements1 Java9ElementsTests#testRootElements2 Java9ElementsTests#testRootElements1Javac Java9ElementsTests#testRootElements2Javac Java9ElementsTests#testAnnotations1 The fact that the first four fail with Javac as well might indicate that the tests are wrong. But I need to investigate more to find out how/why.
PackageInfoTest.test002 fails on 9 because: The method getPackage(String) from the type Package is deprecated
(In reply to Stephan Herrmann from comment #94) > PackageInfoTest.test002 fails on 9 because: > The method getPackage(String) from the type Package is deprecated Oops, there had been a version of the test that expected this warning at 9. Then via bug 527569 this expectation (which was correct IMHO) was replaced by unconditional expectingNoProblems(). Next I found the check for "9" to be bogus, see bug 531362 comment 4, fixed in BETA_JAVA_18_3. With the merge back into master, the two above changes conflicted, and my manual resolution (?) the conditional has been restored - let's await tonights I-build for final result.
New Gerrit change created: https://git.eclipse.org/r/122293
Gerrit change https://git.eclipse.org/r/122293 was merged to [master]. Commit: http://git.eclipse.org/c/jdt/eclipse.jdt.core.git/commit/?id=8e0bec59dc8c471166bf3dc2a5cd34aafc82599a
We still have test issues to address. Will continue through RC1.
We will continue the work through 4.9.
Failure in org.eclipse.jdt.apt.pluggable.tests.FilerTests.testCreateClass1 - Seen in gerrit-jenkins (#1078, #1080 and #1081) thus blocking contributions via gerrit. - Some production builds: http://download.eclipse.org/eclipse/downloads/drops4/I20180727-2000/testresults/html/org.eclipse.jdt.apt.pluggable.tests_ep49I-unit-mac64_macosx.cocoa.x86_64_8.0.html http://download.eclipse.org/eclipse/downloads/drops4/I20180727-2000/testresults/html/org.eclipse.jdt.apt.pluggable.tests_ep49I-unit-cen64-gtk3_linux.gtk.x86_64_8.0.html - Reproduced also locally (on linux). Error is: junit.framework.AssertionFailedError: file should have been overwritten at junit.framework.Assert.fail(Assert.java:57) at junit.framework.Assert.assertTrue(Assert.java:22) at junit.framework.TestCase.assertTrue(TestCase.java:192) at org.eclipse.jdt.apt.pluggable.tests.FilerTests.testCreateClass1(FilerTests.java:399)
(In reply to Stephan Herrmann from comment #100) > Failure in org.eclipse.jdt.apt.pluggable.tests.FilerTests.testCreateClass1 > > - Seen in gerrit-jenkins (#1078, #1080 and #1081) thus blocking > contributions via gerrit. > > - Some production builds: > > http://download.eclipse.org/eclipse/downloads/drops4/I20180727-2000/ > testresults/html/org.eclipse.jdt.apt.pluggable.tests_ep49I-unit-mac64_macosx. > cocoa.x86_64_8.0.html > > http://download.eclipse.org/eclipse/downloads/drops4/I20180727-2000/ > testresults/html/org.eclipse.jdt.apt.pluggable.tests_ep49I-unit-cen64- > gtk3_linux.gtk.x86_64_8.0.html > > - Reproduced also locally (on linux). > > Error is: > > junit.framework.AssertionFailedError: file should have been overwritten > at junit.framework.Assert.fail(Assert.java:57) > at junit.framework.Assert.assertTrue(Assert.java:22) > at junit.framework.TestCase.assertTrue(TestCase.java:192) > at > org.eclipse.jdt.apt.pluggable.tests.FilerTests.testCreateClass1(FilerTests. > java:399) At a second look this has nothing to do with Java 9, so I filed bug 537459.
This bug hasn't had any activity in quite some time. Maybe the problem got resolved, was a duplicate of something else, or became less pressing for some reason - or maybe it's still relevant but just hasn't been looked at yet. If you have further information on the current state of the bug, please add it. The information can be, for example, that the problem still occurs, that you still want the feature, that more information is needed, or that the bug is (for whatever reason) no longer relevant. -- The automated Eclipse Genie.