incubator-odf-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Devin Han <>
Subject Re: My testing notes on RC2
Date Wed, 07 Dec 2011 10:24:22 GMT
2011/12/7 Rob Weir <>

> I tested on two machines:
> 1) Ubuntu 11 running OpenJDK Runtime Environment (IcedTea6 1.11pre)
> (6b23~pre11-0ubuntu1.11.10) and Maven 2.2.1 (rdebian-6)
> 2) Windows XP running Java(TM) SE Runtime Environment (build
> 1.6.0_29-b11) and Maven 3.0.3
> For each environment I downloaded the "0.5-incubating-rc2" tag and
> tried a complete build: via "mvn clean package" from the root.  I also
> download the RC2 source ZIP file and did the same.
> This is what I'm seeing:
> On Linux building from tag, it ends with a unit test failure in the Simple
> API:
> Here is the stack trace:
> -------------------------------------------------------------------------------
> Test set: org.odftoolkit.simple.meta.OfficeMetaTest
> -------------------------------------------------------------------------------
> Tests run: 29, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: 0.682
> sec <<< FAILURE!
> testGetLanguage(org.odftoolkit.simple.meta.OfficeMetaTest)  Time
> elapsed: 0.021 sec  <<< FAILURE!
> junit.framework.ComparisonFailure: null expected:<[rob]> but was:<[en]>
>        at junit.framework.Assert.assertEquals(
>        at junit.framework.Assert.assertEquals(
>        at
> org.odftoolkit.simple.meta.OfficeMetaTest.testGetLanguage(
> This is a functional that gave us trouble in RC1 as well.  IMHO, this
> is not a critical function, so maybe we can disable that test and
> document the problem in the release notes as  a "known issue", with a
> link to a JIRA issue.

I make a mistake here. The test case should get the value of
“user.language" not "". But it should not  failed... As the value
doesn't influence the result, so Oliver get the same result with me. I have
to say your machine is super smart ;)
Anyway , it's my fault and will fixed it in RC3.

> If I disable that test case then the build completes successfully and
> I see all of the generated JAR's.
> This was running the clean and install targets.  We should probably
> document what is needed for generate the release ZIP files.  This
> probably doesn't need to me in the release itself.  But it would be
> great if you could put together some instructions on the wiki for how
> to make a release, based on what you did here.

Agree, I am maintaining a notes about this issue, not only include the
technical content but also the apache release policy for incubator project.
IMHO, the document about how to do an incubator release on Apache website
is too tedious and disordered. I have to jump from one link to another, and
We need a simple and clear one. I will share my notes, when we finish the
first release. Then we can discussion and summary a formal file and
contribute back to Apache.

On the other hand, about our website, I think we have more work to do. Now,
most of its content is migrated from ODFToolkit Union directly, although I
have updated most of link. But I think we need to update and refactor the
content and make it looks like an integration. For example, we need a
download page to list all of the compnents's download links. Each component
can has its own document, but download page should be only one.

Any other guys volunteer to do this work?  IMHO, we need more voices from
other committers, not only Rob and me ;)

> On the Windows build I end on the ODFDOM test with this error:
> Running org.odftoolkit.odfdom.doc.table.TableCellTest
> org.apache.maven.surefire.booter.SurefireExecutionException: Java heap
> space; nested exception is java.lang.OutOfMemoryError: Java heap space
> java.lang.OutOfMemoryError: Java heap space
> I have a 512MB heap, so it is odd to fail in a test case.  I repeated
> with MAVEN-OPTS="-Xmx1025m".  I get the same error.  I tried debugging
> it.  It looks like I'm getting the error in testSetGetFormat(), in the
> 2nd try {} block, when calling
> OdfSpreadsheetDocument.newSpreadsheetDocument().
> I'm seeing weird things going on.  It might be due to some recursion
> issues, or some form of interference with the setUp() function. If you
> look at the setUp() function you see it is also calling static
> functions in OdfSpreadsheetDocument.  And I see setUp called twice
> (but not return) before the call to newSpreadsheetDocument() gives the
> memory exception.  Something weird.  But if it is passing for you and
> others, then it could be an issue with by JVM.  But that test function
> is so long and complex that I don't know.  I think the test code
> should be written to be very simple, so we can be sure it is not a
> source of bugs ;-)

I can volunteer to do it.
But, it's a test case belongs to ODFDOM DOC API, which will be moved in the
next version. Maybe it isn't worth to do it.

> The above tests were done on the RC2 tagged files in SVN.  I also
> repeated the tests with the
> When I do this, nothing is built.  Maven returns after 7 seconds.  No
> errors.  I see this:
> [INFO] Scanning for projects...
> [INFO]
> ------------------------------------------------------------------------
> [INFO] Building Apache ODF Toolkit
> [INFO]    task-segment: [clean, install]
> [INFO]
> ------------------------------------------------------------------------
> [INFO] [clean:clean {execution: default-clean}]
> [INFO] [remote-resources:process {execution: default}]
> [INFO] [site:attach-descriptor {execution: default-attach-descriptor}]
> [INFO] [install:install {execution: default-install}]
> [INFO] Installing /home/rob/odftoolkit-0.5-incubating-rc2/pom.xml to
> /home/rob/.m2/repository/org/apache/odftoolkit/odftoolkit/0.5-incubating-rc2/odftoolkit-0.5-incubating-rc2.pom
> [INFO]
> ------------------------------------------------------------------------
> [INFO]
> ------------------------------------------------------------------------
> [INFO] Total time: 7 seconds
> [INFO] Finished at: Tue Dec 06 15:04:44 EST 2011
> [INFO] Final Memory: 25M/168M
> [INFO]
> ------------------------------------------------------------------------
> The reason is I commeted the following content for a faster process when
upload the binary  and source code packages


> If instead of build from the package root I go to each component and
> "mvn clean install" from the subdir, then it works, except for the
> testGetLanguage() test case mentioned earlier.  If I disable that test
> case then everything else builds OK.
> I also took a quick look at the binary package,  I noticed it was
> larger than the source release, which I thought was odd.  It looks
> like you are including the source ZIP file inside of the binary ZIP.
> That makes is 67MB.

Besides the source ZIP file, ODFDOM javadoc is also with a big size, 20MB.
I think we can remove all of the "with-dependences" jars and the source ZIP
file to make the binary jar smaller. But ODFDOM javadoc maybe must
included, as we have discussed in post of "Directory structure for our
initial release <>"[1], we
should include "docs" directory in the binary package and javadoc is

/odf-0.5/docs (Initially just JavaDoc. We could generate this cross-module,
and have a single JavaDoc site, or produce different sets for each
/odf-0.5/lib (dependency libraries, not all users are using Maven, so we
should probably bundle xerces, xalan, junit, etc.)
/odf-0.5/odf*.jar (one JAR per component)
/odf-0.5/RELEASE-NOTES.txt (consolidated release notes for the project)


> I really like your new README.txt file.  It has a lot of good info.
> -Rob


  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message