Mailing-List: contact user-help@uima.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@uima.apache.org
Content-Type: multipart/alternative;
 boundary="Apple-Mail=_90666511-F546-439D-ADD8-210A1FD93BC0"
Mime-Version: 1.0 (Mac OS X Mail 8.2 \(2070.6\))
Subject: Re: NPE when trying to get casProcessors from CPE description
From: Martin Wunderlich <martin_wu@gmx.net>
In-Reply-To: <29633835-E0D7-44A1-90AA-CD1B6159BAEF@gmx.net>
Date: Sun, 12 Jul 2015 20:49:00 +0200
Cc: dkpro-tc-users@googlegroups.com
Message-Id: <CAB67763-70CE-4B28-A21E-79BD4BF20B97@gmx.net>
References: <5E86035F-5381-426A-A221-835CA6C85FB6@gmx.net>
 <DBE7CD31-2B8F-4BD2-9CAD-19242EE4DFDA@apache.org>
 <601FBEB5-6F01-484C-908A-CF62C5543DA6@gmx.net>
 <29633835-E0D7-44A1-90AA-CD1B6159BAEF@gmx.net>
To: user@uima.apache.org

--Apple-Mail=_90666511-F546-439D-ADD8-210A1FD93BC0
Content-Transfer-Encoding: quoted-printable
Content-Type: text/plain;
	charset=windows-1252

Hi,=20

I just want to briefly reveal how this situation has been resolved.=20

First, Richard was so kind to contact me via chat to check the Maven =
dependencies. We discovered that two different versions of DKPro Lab =
were referenced (one of them rather indirectly, from the similarity =
project). Richard quickly resolved this with a commit and I re-installed =
the snapshot version locally. However, this didn=92t fix this situation =
listed below. But kudos to Richard who helped out and made me feel less =
lonely on this hot summer sunday.=20

As a next step, I tried to isolate the problem. So, I cloned the TC =
github repo, imported that into a new Eclipse workspace and ran the =
example there. This worked. So, I figured it might be something in my =
particular workspace arrangement. I had my forked TC project loaded =
there, which I now set to the master branch of the original repo=92s =
snapshot version. This didn=92t help either. Then, I re-installed the =
newly cloned version using maven install, removed the TC project from =
the eclipse workspace, imported only the examples project and ran the =
example again. Now it worked. I went back to my actual project and - lo =
and behold - the pair-wise classification now worked, too.=20

I still don=92t know what caused the problem. It might have been some of =
my code changes or config changes to the forked copy of the project. =
Anyway, I am documenting this here in case someone else runs into a =
similar situation.=20

Cheers,=20

Martin
=20

> Am 12.07.2015 um 16:39 schrieb Martin Wunderlich <martin_wu@gmx.net>:
>=20
> As a side-note*, the problem does not occur when I run the experiment =
from the branch of the 0.7.0 tag version of DKPro TC. In that version, =
PairTwentyNewsgroupsDemo runs fine without the NPE. Not sure, if this =
observation adds to the clarification or rather to the confusion.=20
>=20
> Cheers,=20
>=20
> Martin
>=20
> * (and we should probably move this conversation to the DKpro list, =
which I am putting on CC now, so feel free to respond there)
> =20
>=20
>> Am 12.07.2015 um 13:23 schrieb Martin Wunderlich <martin_wu@gmx.net =
<mailto:martin_wu@gmx.net>>:
>>=20
>> Hi Richard,=20
>>=20
>> I have based this experiment on the PairTwentyNewsgroupsDemo from the =
DKPro examples. The CPE would be build during the experiment execution, =
I guess.=20
>>=20
>> I have just tried to re-run the example, but now I am getting the =
same exception:=20
>> (using the 0.8.0 snapshort version of TC).=20
>>=20
>> 2015-07-12 11:01:31 DEBUG [main] (DefaultLoggingService) - =
[MetaInfoTask-TwentyNewsgroupsTrainTest-966ab1a7-2874-11e5-b53c-37fcf6476d=
26] Problem stack trace:
>> java.lang.NullPointerException
>> 	at =
org.apache.uima.fit.cpe.CpeBuilder.createCpe(CpeBuilder.java:159)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.uima.engine.cpe.CpeExecutionEngine.run(CpeExe=
cutionEngine.java:93)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.task.impl.BatchTask.runNewExecution(BatchTask=
.java:350)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.task.impl.BatchTask.executeConfiguration(Batc=
hTask.java:255)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.task.impl.BatchTask.execute(BatchTask.java:18=
5)
>> 	at =
de.tudarmstadt.ukp.dkpro.tc.ml.ExperimentTrainTest.execute(ExperimentTrain=
Test.java:91)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.engine.impl.ExecutableTaskEngine.run(Executab=
leTaskEngine.java:55)
>> 	at =
de.tudarmstadt.ukp.dkpro.lab.engine.impl.DefaultTaskExecutionService.run(D=
efaultTaskExecutionService.java:48)
>> 	at de.tudarmstadt.ukp.dkpro.lab.Lab.run(Lab.java:97)
>> 	at =
de.tudarmstadt.ukp.dkpro.tc.examples.single.pair.PairTwentyNewsgroupsDemo.=
runTrainTest(PairTwentyNewsgroupsDemo.java:147)
>> 	at =
de.tudarmstadt.ukp.dkpro.tc.examples.single.pair.PairTwentyNewsgroupsDemo.=
main(PairTwentyNewsgroupsDemo.java:84)
>>=20
>> Cheers,=20
>>=20
>> Martin
>> =20
>>=20
>>=20
>>> Am 11.07.2015 um 19:10 schrieb Richard Eckart de Castilho =
<rec@apache.org <mailto:rec@apache.org>>:
>>>=20
>>> Hi,
>>>=20
>>> how to you build your CPE? The CAS processors should be set during =
setAnalysisEngine(...).
>>>=20
>>> Cheers,
>>>=20
>>> -- Richard=20
>>>=20
>>> On 11.07.2015, at 10:18, Martin Wunderlich <martin_wu@gmx.net =
<mailto:martin_wu@gmx.net>> wrote:
>>>=20
>>>> Hi  all,=20
>>>>=20
>>>> I am  trying to run a pair-wise classification experiment using the =
DKPro TC framework. The execution fails during the meta-info stage due =
to a NullPointerException that originates in the CpeBuilder class from =
UIMAfit v.2.1.0 (hence the post to this list and not the DKpro list). I =
had a closer look at the code and the NPE is triggered at the following =
line in the CpeBuilder:=20
>>>>=20
>>>> =
cpeDesc.getCpeCasProcessors().setPoolSize(maxProcessingUnitThreadCount + =
2);
>>>>=20
>>>> The underlying root cause is that the casProcessors object returned =
by the call to cpeDesc is Null, but I don=92t understand why. The =
cpeDesc lists a number of fields as having Null values:=20
>>>>=20
>>>> org.apache.uima.collection.impl.metadata.cpe.CpeDescriptionImpl:=20
>>>> casInitializer =3D NULL
>>>> configParameterSettings =3D NULL
>>>> configurationParameterSettings =3D NULL
>>>> import =3D NULL
>>>> sofaNameMappings =3D NULL
>>>> configurationParameterSettings =3D NULL
>>>> cpeCasProcessors =3D NULL
>>>> outputQueue =3D NULL
>>>> startAt =3D NULL
>>>> startingEntityId =3D NULL
>>>>=20
>>>>=20
>>>> Unfortunately, I am not too familiar with the inner workings of the =
UIMA(fit) framework so I am not sure how, why and where these fields =
were set to Null.=20
>>>> Could someone perhaps point me in the right direction where I =
should dig deeper to identify the cause for this problem?=20
>>>>=20
>>>> Thanks a lot.=20
>>>>=20
>>>> Cheers,=20
>>>>=20
>>>> Martin
>>>=20
>>=20
>=20


--Apple-Mail=_90666511-F546-439D-ADD8-210A1FD93BC0--