In addition, there's a thread leak in 'forever' mode.

Patch adds a test for enwiki-content-source and fixes both the last doc problem and the thread leak.

Patch looks good ! Nice catch. Few minor (styling) comments, feel free to ignore:

• In EnwikiCSTest.Stringable, I would make docs 'final'.

• I would change assertNoMoreDataException like this:

private void assertNoMoreDataException(EnwikiContentSource stdm) throws Exception {
  try {
    stdm.getNextDocData(null);
    fail("Expecting NoMoreDataException");
  } catch (NoMoreDataException e) {
    // expected
  }
}

Thanks for the review Shai, changed as you suggested and committed (while jira was down...)

[trunk commit] Doron Cohen
http://svn.apache.org/viewvc?view=revision&revision=1417788

LUCENE-4588: EnwikiContentSource fixes: lost last wiki doc, thread leak in 'forever' mode.

Two more commits to trunk (uncaught by bot due to incorrect message format):

• r1417871 – LUCENE-4588 (cont): (EnwikiContentSource fixes) avoid using the forbidden
  StringBufferInputStream..
• r1417921 – LUCENE-4588 (cont): simplify test input stream crration.

[branch_4x commit] Doron Cohen
http://svn.apache.org/viewvc?view=revision&revision=1418892

LUCENE-4588: merge from trunk: EnwikiContentSource silently swallows the last wiki doc.

[branch_4x commit] Doron Cohen
http://svn.apache.org/viewvc?view=revision&revision=1418894

LUCENE-4588: (cont) add missing property svn:eol-style.

Fixed.

As a side note, merging benchmark changes to 4x is so much easier than it used to be in 3x, now that trunk and branch are structured the same! Now if only 'precommit' would run 60 times faster (that would be 12 seconds here)... wouldn't that be great?

[branch_4x commit] Doron Cohen
http://svn.apache.org/viewvc?view=revision&revision=1418894

LUCENE-4588: (cont) add missing property svn:eol-style.

[branch_4x commit] Doron Cohen
http://svn.apache.org/viewvc?view=revision&revision=1418892

LUCENE-4588: merge from trunk: EnwikiContentSource silently swallows the last wiki doc.