Currently, if you specify an empty infoset as the expected result (e.g. <tdml:infoset><tdml:dfdlInfoset/></tdml:infoset>), you get the following exception:

ois/ncsa/daffodil/section06/namespaces/multi_B_01.dfdl.xsd.]
[info] - test_multifile_choice_02b *** FAILED ***
[info] java.lang.IndexOutOfBoundsException: 0
[info] at scala.collection.LinearSeqOptimized$class.apply(LinearSeqOptimized.scala:51)
[info] at scala.collection.immutable.List.apply(List.scala:76)
[info] at edu.illinois.ncsa.daffodil.tdml.DFDLInfoset.contents(TDMLRunner.scala:893)
[info] at edu.illinois.ncsa.daffodil.tdml.Infoset.contents(TDMLRunner.scala:888)
[info] at edu.illinois.ncsa.daffodil.tdml.ParserTestCase.verifyParseInfoset(TDMLRunner.scala:408)
[info] at edu.illinois.ncsa.daffodil.tdml.ParserTestCase.runParseExpectSuccess(TDMLRunner.scala:494)
[info] at edu.illinois.ncsa.daffodil.tdml.ParserTestCase.runProcessor(TDMLRunner.scala:372)
[info] at edu.illinois.ncsa.daffodil.tdml.TestCase.run(TDMLRunner.scala:326)
[info] at edu.illinois.ncsa.daffodil.tdml.DFDLTestSuite.runOneTest(TDMLRunner.scala:198)
[info] at edu.illinois.ncsa.daffodil.section06.namespaces.TestNamespaces.test_multifile_choice_02b(TestNamespaces.scala:59)

I think the TDML runner should just compare the proper result with the empty set and provide this to the user so they can adjust the expected infoset accordingly, instead of allowing them to think there is an issue with the parsing.