Bugzilla – Full Text Bug Listing
| Summary: | org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest.testJspFileExistence fails for 2022-03 | ||
|---|---|---|---|
| Product: | [Modeling] MoDisco | Reporter: | Ed Willink <ed> |
| Component: | Main | Assignee: | Project Inbox <modisco.web-inbox> |
| Status: | RESOLVED FIXED | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | ||
| Version: | unspecified | ||
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows 10 | ||
| Whiteboard: | |||
| Bug Depends on: | 479450 | ||
| Bug Blocks: | |||
|
Description
Ed Willink
(In reply to Ed Willink from comment #0) > Fails to reproduce iin my workspace for which the content is consistent. Attempting to use workspace interactively falls foul of Bug 579441, but once fixed, running org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest as a JUnit plugin test works fine. So: 2022-03 JUnit plugin Windows => ok 2022-03++ I-build Tycho/m2e Windows => ok 2022-03++ I-build Tycho/Jenkins Linux => bad Looks like a Linux/Windows difference. Need to develop stronger debugging interactively so that it can be run in branch-tests. (In reply to Ed Willink from comment #0) > <%-- Comment JSP Element --%> <%@ include name="value"%> <%-- Comment JSP ... > include name="value"%> This is some text to test the jsp:text function Oops not /*view*/ with /*view*/ <jsp:element name="value" name1="value1" /> <jsp:element name="value" name1="value1" > <%-- Comment JSP Element --%> <%@ include name="value"%> </jsp:element> ... <jsp:text>This is some text to test the jsp:text function </jsp:text> --- some whitespsce differences are obvious but they occur in an ok run in my workspace. After a bt of debugging, it turns out that the 'filecomparator' cannot cope with the whitespace anomalies and so has been trivialized to just a file name comparison. How can that be going wrong? ?? the ".." file ?? but then how did it ever work ?? (In reply to Ed Willink from comment #1) > Looks like a Linux/Windows difference. No it does reproduc locally, but test failures do not fail the build so it loks like success. On Jenkins the attempt to find the testresults provides a gone-unstable fall-back. With extra debugging the failure is Original Jsp files location : C:\Users\ed\modisco-master2\git\org.eclipse.modisco\org.eclipse.modisco.jee.jsp.generation.tests\target\work\data\Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest comparison of Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest and JspOutput folders Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest and JspOutput do not have the same number of children (2, 1) folders Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest and JspOutput are not equal. Tests run: 1, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 6.192 s <<< FAILURE! - in org.eclipse.modisco.jee.jsp.generation.tests.AllTestsInUIThread testJspFileExistence(org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest) Time elapsed: 6.161 s <<< FAILURE! java.lang.AssertionError: Generated JSP files differ from original files Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest has an extra /org.eclipse.modisco.jee.jsp.generation.tests/target/work/data/Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest/.settings/org.eclipse.core.resources.prefs whose content eclipse.preferences.version=1 encoding/<project>=Cp1252 does not correspond to project settings. (In reply to Ed Willink from comment #3) > Jsp.org.eclipse.modisco.jee.jsp.generation.tests.DiffGeneratedJspTest has an > extra ... > .settings/org.eclipse.core.resources.prefs A bad m2e / Tycho handling of project resources copy would seem like the obvious culprit, but we're still using Tycho 2.1.0 so perhaps the only possibility is the platform I-build. Ah! Bug 479450 is an increasingly contentious platform chanfe. The .settings/org.eclipse.core.resources.prefs is created by writeEncodingAfterOpen(monitor) in Project.open(). (In reply to Ed Willink from comment #4) > Ah! Bug 479450 is an increasingly contentious platform chanfe. But actually a good one. Ensuring that the MoDisco test ignores .settings is easy enough. Pushed to master. |