[
https://issues.apache.org/jira/browse/OPENNLP-669?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
James Kosin updated OPENNLP-669:
--------------------------------
Attachment: test_eventstream_java_1.8.0.txt
test_eventstream_java_1.7.0_51.txt
The event streams are identical between the two versions. (see attached)
I'm going to dig into the training and try to determine what happens there.
Maybe something isn't being done right or there is a bug to file with Java 8.
It is a pre-released version before the final release to everyone.
> Java 1.8 incompatibility and Build failures
> -------------------------------------------
>
> Key: OPENNLP-669
> URL: https://issues.apache.org/jira/browse/OPENNLP-669
> Project: OpenNLP
> Issue Type: Improvement
> Components: Build, Packaging and Test, Documentation
> Affects Versions: 1.6.0
> Environment: Windows 8.1 Java 1.8
> Reporter: James Kosin
> Assignee: James Kosin
> Labels: java1.8, java8
> Fix For: 1.6.1
>
> Attachments: java_test_1.7.0_51.txt, java_test_1.8.0.txt,
> test_eventstream_java_1.7.0_51.txt, test_eventstream_java_1.8.0.txt
>
> Original Estimate: 504h
> Remaining Estimate: 504h
>
> I just wanted to let everyone know, the OpenNLP project is not compatible
> with the new Java 1.8.
> I get two (2) issues:
> 1) Fails test:
> Failed tests:
> testPerceptronOnPrepAttachDataWithStepSizeDecrease(opennlp.tools
> .ml.perceptron.PerceptronPrepAttachTest):
> expected:<0.7756870512503095> but was:
> <0.7766773953948998>
> 2) Java document generation fails miserably.
> [ERROR] * <li>.bin --> the file is binary
> [ERROR] ^
> [ERROR]
> C:\Users\jkosin\Documents\Projects\OpenNLP\opennlp\opennlp-tools\src\mai
> n\java\opennlp\tools\ml\perceptron\SuffixSensitivePerceptronModelWriter.java:41:
> error: bad use of '>'
> [ERROR] * <li>.bin --> the file is binary
> [ERROR] ^
> [ERROR]
> C:\Users\jkosin\Documents\Projects\OpenNLP\opennlp\opennlp-tools\src\mai
> n\java\opennlp\tools\postag\POSModel.java:64: error: reference not found
> The second for javadoc errors are probably due to more stringent checking in
> javadoc.
> The first... is a bit disconcerting and troubling. I'm going to check into
> this; because this error shouldn't be happening.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
