I get the same kind of errors on my Linux box. I don’t understand why the error under Java 1.4 does not occur when xercesImpl.jar is added to the classpath, since AFAIK we don’t trigger the endorsed mechanism, so the default SAX parser implementation shipped with Java should not be overridden.
I guess the simplest is to keep with the status quo, or with your solution #2 if you feel like implementing it. However, I would put only serializer and xercesImpl in the endorsed/ directory, since like I explained [1] I’d consider the other two as regular dependencies (classes not available in the standard library). (That said, newbies will probably download the binaries instead of having to fiddle with the source files.) Vincent [1] http://mail-archives.apache.org/mod_mbox/xmlgraphics-fop-dev/200801.mbox/[EMAIL PROTECTED] Jeremias Maerki wrote: > I've done some experiments based on our discussion. I removed Xerces and > serializer.jar and moved Xalan-J to a test/lib directory and made it > available only to the test code. Building FOP like this is no problem > but running the test suite on a basic Sun JDK 1.4.2_16 (with no replaced > XML JARs) fails: > > Testsuite: org.apache.fop.layoutengine.LayoutEngineTestSuite > Tests run: 7, Failures: 0, Errors: 1, Time elapsed: 3.547 sec > ------------- Standard Error ----------------- > <snip/> > ------------- ---------------- --------------- > > Testcase: block_hyphenation-ladder-count.xml took 1.375 sec > Testcase: block_hyphenation_kerning.xml took 0.094 sec > Testcase: block_hyphenation_no-wrap.xml took 0.031 sec > Testcase: block_uax14_linebreaking_hyph.xml took 0.016 sec > Caused an ERROR > Dokumentwurzelelement fehlt > org.xml.sax.SAXParseException: Dokumentwurzelelement fehlt > at org.apache.crimson.parser.Parser2.fatal(Parser2.java:3376) > at org.apache.crimson.parser.Parser2.fatal(Parser2.java:3364) > at org.apache.crimson.parser.Parser2.parseInternal(Parser2.java:668) > at org.apache.crimson.parser.Parser2.parse(Parser2.java:337) > at org.apache.crimson.parser.XMLReaderImpl.parse(XMLReaderImpl.java:448) > at > org.apache.crimson.jaxp.DocumentBuilderImpl.parse(DocumentBuilderImpl.java:185) > at javax.xml.parsers.DocumentBuilder.parse(DocumentBuilder.java:151) > at > org.apache.fop.layoutengine.LayoutEngineTester.runTest(LayoutEngineTester.java:139) > at > org.apache.fop.layoutengine.LayoutEngineTestSuite$LayoutEngineTestCase.testMain(LayoutEngineTestSuite.java:214) > at > org.apache.fop.layoutengine.LayoutEngineTestSuite$1.runTest(LayoutEngineTestSuite.java:193) > > Testcase: footnote_in_inline.xml took 0.031 sec > Testcase: inline_border_padding_hyphenate.xml took 1.297 sec > Testcase: inline_border_padding_hyphenate_de.xml took 0.672 sec > > > "Dokumentwurzelelement fehlt" translates to "Document root element > is missing". Probably a bug in Crimson. > > Even worse on Sun JDK 1.5.0_14 and 6.0_03: > > [junit] Testcase: > test_fonts_directory_recursive.xconf(org.apache.fop.config.FontsDirectoryRecursiveTestCase): > Caused an ERROR > [junit] org/apache/xml/serializer/TreeWalker > [junit] java.lang.NoClassDefFoundError: > org/apache/xml/serializer/TreeWalker > [junit] at > org.apache.xalan.processor.TransformerFactoryImpl.newTransformer(TransformerFactoryImpl.java:823) > [junit] at > org.apache.fop.render.pdf.BasePDFTestCase.convertFO(BasePDFTestCase.java:88) > [junit] at > org.apache.fop.config.BaseUserConfigTestCase.convertFO(BaseUserConfigTestCase.java:71) > [junit] at > org.apache.fop.config.BaseConstructiveUserConfigTestCase.testUserConfig(BaseConstructiveUserConfigTestCase.java:38) > [junit] > > and... > > junit-layout-standard: > [echo] Running standard layout engine tests > [junit] Testsuite: org.apache.fop.layoutengine.LayoutEngineTestSuite > [junit] Tests run: 1, Failures: 0, Errors: 1, Time elapsed: 0 sec > [junit] > [junit] Null Test: Caused an ERROR > [junit] null > [junit] java.lang.reflect.InvocationTargetException > [junit] Caused by: java.lang.NoClassDefFoundError: > org/apache/xml/serializer/ExtendedContentHandler > [junit] at > org.apache.xalan.processor.XSLTSchema.build(XSLTSchema.java:325) > [junit] at > org.apache.xalan.processor.XSLTSchema.<init>(XSLTSchema.java:72) > [junit] at > org.apache.xalan.processor.StylesheetHandler.<init>(StylesheetHandler.java:1290) > [junit] at > org.apache.xalan.processor.TransformerFactoryImpl.newTemplatesHandler(TransformerFactoryImpl.java:376) > [junit] at > org.apache.xalan.processor.TransformerFactoryImpl.newTemplates(TransformerFactoryImpl.java:867) > [junit] at > org.apache.xalan.processor.TransformerFactoryImpl.newTransformer(TransformerFactoryImpl.java:776) > [junit] at > org.apache.fop.layoutengine.LayoutEngineTestSuite.readDisabledTestcases(LayoutEngineTestSuite.java:71) > [junit] at > org.apache.fop.layoutengine.LayoutEngineTestSuite.decorateWithDisabledList(LayoutEngineTestSuite.java:122) > [junit] at > org.apache.fop.layoutengine.LayoutEngineTestSuite.getTestFiles(LayoutEngineTestSuite.java:146) > [junit] at > org.apache.fop.layoutengine.LayoutEngineTestSuite.suite(LayoutEngineTestSuite.java:174) > [junit] > [junit] > [junit] Test org.apache.fop.layoutengine.LayoutEngineTestSuite FAILED > > which may a side effect of our using the ASF-built xml-apis.jar which > has a fallback to org.apache.xalan.processor.TransformerFactoryImpl for > the TransformerFactory. Why the fallback is triggered eludes me. > > At least, restoring serializer.jar to the test/lib directory lets the > test suite pass on Java 5 and 6. So we're back to only one JAR that can > be removed and that would actually be useful in a Java 1.4 context which > we'll use to build the release. > > I want FOP to build out-of-the-box without the user having to fiddle > around with his system (if he has Ant properly installed which can be > assumed today). A newbie needs a "quick success" when he downloads FOP > for the first time. I usually throw new software I try out away if I > cannot build from the sources within 10 minutes. Usually happens with > Maven-built software but not exclusively. Basically, I'm back with my > original solution 2 (<fop-root>/lib/endorsed, no JAR removal) or status > quo. > > Did I miss anything? > > > Jeremias Maerki > -- Vincent Hennebert Anyware Technologies http://people.apache.org/~vhennebert http://www.anyware-tech.com Apache FOP Committer FOP Development/Consulting
