fixed in trunk. Thanks for testing :-) -Marshall
On 10/11/2017 9:50 AM, Marshall Schor wrote: > I think I found the cause: a faulty implementation of load(asInputStream, > aCAS, > typeSystem). > > Looking at the code, the 3rd argument is not used! It looks like something > happened when migrating from V2 -> V3 code base, because it's OK in the v2 > code > base. Maybe I can blame the cat (she sometimes jumps up onto my desk, landing > on my keyboard, and randomly pressing keys...) :-) > > Apologies, and thank goodness for testing... > > -Marshall > > On 10/10/2017 4:45 PM, Richard Eckart de Castilho wrote: >> On 10.10.2017, at 22:03, Richard Eckart de Castilho <r...@apache.org> wrote: >>> On 09.10.2017, at 16:35, Marshall Schor <m...@schor.com> wrote: >>>> I think this will need a test case - can you provide one? >>> I can see the extended error message and will try to produce a reduced unit >>> test. >>> >>> org.apache.uima.cas.CASRuntimeException: Deserializing Compressed Form 6, a >>> type code: 71 has no corresponding type. currentFsId: 1 nbrFSs: 0 >>> nextFsAddr: 1 >>> at >>> org.apache.uima.cas.impl.BinaryCasSerDes6.deserializeAfterVersion(BinaryCasSerDes6.java:1812) >>> at >>> org.apache.uima.cas.impl.BinaryCasSerDes.reinit(BinaryCasSerDes.java:594) >>> at org.apache.uima.util.CasIOUtils.load(CasIOUtils.java:382) >>> at org.apache.uima.util.CasIOUtils.load(CasIOUtils.java:344) >>> ... >> This seems to happen only when CasIOUtils.load(casInputStream, aCAS, >> typeSystem) is used. >> I tried to set up a minimal test which serializes a CAS into one byte-array >> for the contents >> and a second one for the type system, then deserializes this into a new CAS >> using the >> method above. However, this doesn't seem to trigger the problem. So it might >> be related >> to the fact that in the failing DKPro Core test, there are also two >> collection reader components >> and an analysis engine (actually a writing component) involved. >> >> I'll try to expand the test until it fails. If you have any additional >> pointers, please >> let me know. >> >> Cheers, >> >> -- Richard >
