Hello again. I'm sorry if I'm being a bit persistent. I just see this falling
out of view once the weekend arrives. Let know if there is some additional
information I can provide that may offer additional insights or if there is
another forum with which I should be addressing this question.
From: Murphy, Sean P. [RO IS]
Sent: Thursday, January 12, 2017 9:22 AM
Subject: Annotation Index not recognized with Linux
I am seeing some behavior that appears to be specific to our Jenkins build
environment where the base annotation (uima.tcas.Annotation) is not being
sorted correctly. The expected behavior, where tokens are sorted in an
ascending order based on the begin token integer, is occurring on any other
platform (iOS or Windows). I attempted to circumvent this behavior by
utilizing the Indexes Page to add an additional index which uses OHNLP's
ConceptMention as 'kind' 'sorted' and features similar to the built in index
provided, but that did not have the intended result either.
To establish if the added index had any effect I went on to use
reverse sort direction on the ConceptMention annotation, but it didn't appear
to take effect in any environment which leads me to believe I'm not utilizing
it correctly. I'm wondering if the additional index needs to be made on a
primitive aggregate rather than at the AE level as I'm attempting. It is not
clear by the documentation whether the getAnnotationIndex() method has access
to the sorted index of if another method (such as FSIndexRepository) would be
necessary to produce the iterator which honors the indexing. I've tried both
with no apparent effect.
So I guess I'm asking two different questions:
1) Are there issues with Linux utilizing the built in index annotation?
2) How does one go about implementing the indexes pages to provide a
custom handling of sorting indexes?