[
https://issues.apache.org/jira/browse/UIMA-4565?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14711697#comment-14711697
]
Marshall Schor commented on UIMA-4565:
--------------------------------------
I agree. The logic here: AnnotationBase has subject-of-analysis refs;
Annotation is a subclass where the spans are denoted by begin/end pairs (ints);
other kinds of subclasses of AnnotationBase have subject-of-analysis refs like
"images" with pixels, and spans are more naturally (perhaps) denoted by 2
2-element sets of x,y coordinates (assuming rectangular areas). These others
are not (of course) "built-in"... So the sofa feat should be on AnnotationBase.
> CAS.FEATURE_FULL_NAME_SOFA wrong?
> ---------------------------------
>
> Key: UIMA-4565
> URL: https://issues.apache.org/jira/browse/UIMA-4565
> Project: UIMA
> Issue Type: Bug
> Components: Core Java Framework
> Affects Versions: 2.8.1SDK
> Reporter: Richard Eckart de Castilho
> Fix For: 2.8.2SDK
>
>
> The CAS interface has to say this about the *sofa* feature's full name
> {noformat}
> FEATURE_FULL_NAME_SOFA = "uima.tcas.Annotation:sofa";
> {noformat}
> However, when I try to use this to filter out the sofa feature, I don't catch
> it - and the reason seems to be that the *sofa* feature is actually defined as
> {noformat}
> uima.cas.AnnotationBase:sofa
> {noformat}
> ... at least that's what the debugger says.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
