Recent changes to bugs

Document Analyzer ignores preferences

2006-11-10T17:15:04Z

Every time I run the Document Analyzer the character
encoding is set to UTF-8, but I use win-1252. I
looked at the source of
com.ibm.uima.reference_impl.application.docanalyzer.DocumentAnalyzer
and UTF-8 es hardcoded.

Boolean & Integers given default value when exporting CAS

2006-11-09T00:05:22Z

When exporting a CAS to xmi using the UIMA
applications (Document Analyser) we are finding that
Boolean & Integer types are given a default value
whereas Strings are not as they can be set as null.
Is this the expected behaviour?

E.g. I have a typed system named 'Person' with
attributes name:String & validName:Boolean.

Within the AE the 'validName' is not defined as
either a input or output and is not set within the
respective Annotator class. However in the xmi file
generated I am getting a result for both attributes
being 'name' and 'validName'.

If I change the the 'validName' attribute to be a
String, the attribute is not written to the xmi.

Thanks

Evan

UIMA eclipse preference page cannot be loaded under Linux

2006-11-04T11:05:53Z

This is due to a wrong Plugin Descriptor for
com.ibm.uima.desceditor. The property Bundle-ClassPath
points to "TaeConfigurator.jar". While this might work
under Windows, in a Linux environment it does not.

Fix: either rename the jar file to
"TaeConfigurator.jar" or change the descriptor property
(in
$ECLIPSE_HOME/plugins/com.ibm.uima.desceditor/META-INF/MANIFEST.MF)
to "Bundle-ClassPath: TAEconfigurator.jar".

Attached you'll find the modified descriptor file.

CAS serialization is not documented

2006-10-11T22:26:42Z

Class com.ibm.uima.cas.impl.Serialization is not
documented in the JavaDoc.

Since class CAS doesn't implement Serializable,
Serialization is the only way to (efficiently)
transport a CAS Java process to Java process (via RMI
or Jini).

Also, Serialization seems to be missing a method.
There is a "CASSerializer serializeCAS(CAS)" method
but no way to deserialize it. There is a pair that
works on Streams, but there is no "void
deserializeCAS(CAS, CASSerializer)" method.

Secondly, the method serializeCASComplete() takes a
CASMgr, which is also undocumented. Perhaps this
should be modified to take a CAS instead of a CASMgr?

Eclipse plugin failure in Linux

2006-09-21T16:42:26Z

The com.ibm.uima.desceditor_2.0.1 eclipse plugin fails
in Linux because of the case sensitive file system that
Window$ does not have. In the MANIFEST.MF for the
eclipse plugin contains "Bundle-ClassPath:
TaeConfigurator.jar" when in fact the jars name is
"TAEconfigurator.jar". The capitalization causes a
problem when running on a Linux FS.

-Stephen

Can't call process twice on CPE

2006-09-06T02:05:13Z

Hello there,

I've been having lots of fun with UIMA 2.0.0

My problem occurs when I call process() twice on a
CollectionProcessingEngine. The call bombs out with a
next to useless thread exception.

6/09/2006 11:28:39
com.ibm.uima.reference_impl.collection.cpm.BaseCPMImpl
run(515)
SEVERE: java.lang.IllegalThreadStateException

The docs appear to suggest its fine to call process
more than once just not at the same time.

Any help will be appreciated.

2.0 examples use deprecated methods

2006-08-30T04:05:23Z

For example, in
docs\examples\src\com\ibm\uima\examples\RunAE.java,
CollectionReader.setCasInitializer() is called, which
is deprecated.

The examples should show the right way to do things.
In this case, it should show how to use a multi-SOFA
annotator instead of a CAS Initializer. For example,
it might use XmlDetagger annotator.

I'm sure there are other places the examples use
deprecated methods, so a review of that code might be
appropriate to bring it up to the 2.0 APIs.

2.0 examples use deprecated methods

2006-08-30T00:51:19Z

For example, in
docs\examples\src\com\ibm\uima\examples\RunAE.java,
CollectionReader.setCasInitializer() is called, which
is deprecated.

The examples should show the right way to do things.
In this case, it should show how to use a multi-SOFA
annotator instead of a CAS Initializer. For example,
it might use XmlDetagger annotator.

I'm sure there are other places the examples use
deprecated methods, so a review of that code might be
appropriate to bring it up to the 2.0 APIs.

Infinite loop in descriptor editor

2006-08-10T06:12:09Z

In the Component Descriptor Editor, if you are in the
Source tab and make a mistake (for example, an import
path is wrong and the file can't be read), then when
you go to another tab, the editor will pop up an Error
dialog. You click OK, and it pops up another
(identical) Error dialog. Click OK, and the process
repeats without end.

Your only choice is to kill the java.exe process
that's running Eclipse.

NumberFormatException in checkpoint

2006-08-01T04:49:59Z

com.ibm.uima.reference_impl.collection.metadata.
checkpoint.getBatch() can throw a java.lang.
NumberFormatException:

Exception in thread "main" java.lang.
NumberFormatException: null
at java.lang.Integer.parseInt(Unknown Source)
at java.lang.Short.parseShort(Unknown Source)
at java.lang.Short.<init>(Unknown Source)
at com.ibm.etools.xmlschema.beans.ComplexType.
getAttributeShortValue(ComplexType.java:312)
at com.ibm.uima.reference_impl.collection.
metadata.checkpoint.getBatch(checkpoint.java:61)
at com.ibm.uima.reference_impl.collection.
metadata.cpe.CpeCheckpointImpl.
getBatchSize(CpeCheckpointImpl.java:103)