Martin Wiesner created OPENNLP-1702:
---------------------------------------
Summary: BratDocumentStream should process files in bratCorpusDir
deterministically
Key: OPENNLP-1702
URL: https://issues.apache.org/jira/browse/OPENNLP-1702
Project: OpenNLP
Issue Type: Bug
Components: Build, Packaging and Test
Affects Versions: 2.5.3
Reporter: Martin Wiesner
Assignee: Martin Wiesner
Fix For: 2.5.4
With the recent addition of {{BratNameSampleStreamFactoryTest}} via
OPENNLP-1695, it became obvious (Eval test run), that the code in
BratDocumentStream is prone to non-determinism. This stems from the fact that
{{java.util.File#listFiles(..)}} does not guarantee any order of the returned
elements.
A potential fix for achieving determinism again, is to sort the result of
listFiles(..) alphabetically in ASC order.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
