tika-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Chris A. Mattmann (Commented) (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (TIKA-888) NetCDF parser uses Java 6 JAR file and test/compilation fails with Java 1.5, although TIKA is Java 1.5
Date Fri, 30 Mar 2012 14:20:28 GMT

    [ https://issues.apache.org/jira/browse/TIKA-888?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13242368#comment-13242368

Chris A. Mattmann commented on TIKA-888:

Hi Uwe, thanks for reporting this. Unfortunately that version of the NetCDF Java library is
published under 1.6, and it was a long process to get the Jar out onto Maven Central (I actually
published it there, and became part of the NetCDF community ,helping them modernize their
build to Maven, etc. etc.). See: http://www.unidata.ucar.edu/mailing_lists/archives/netcdf-java/2010/msg00129.html
for an example of the discussion.

I'm not sure what the problem is having a 1.6 dependency in Tika. We disable the tests for
NetCDF java if the code isn't being compiled on a 1.6 platform, so all of the Tika code compiles
just fine. And, because of the service provider mechanism that we use to load parser classes,
only external users of the tika-parsers library that want to use NetCDF support for scientific
data formats (NetCDF, OPeNDAP, HDF, etc.) need to worry about the 1.6 requirement. tika-bundle,
and tika-app, (and now tika-server) slurp the jar file up as a dependency, but downstream
users can simply use exclusion tactics (via Ivy or Maven) to not include it in their specific
software projects.

JDK 1.5 is EOL'ed, so eventually we'll all need to be on 1.6 or better (1.7+) anyways, so
I'd favor simply revisiting the discussion I'm about to reference below before changing anything
in Tika land or working to try and publish a new NetCDF jar that's 1.5 compat. Jukka and I
discussed this in Tika land a while back (November 2010), see here: http://s.apache.org/6ij
when we were finding Jenkins issues for Tika builds.

Can't Solr simply use an Ivy exclusion (since I see you guys are moving to Ivy)?
> NetCDF parser uses Java 6 JAR file and test/compilation fails with Java 1.5, although
TIKA is Java 1.5
> ------------------------------------------------------------------------------------------------------
>                 Key: TIKA-888
>                 URL: https://issues.apache.org/jira/browse/TIKA-888
>             Project: Tika
>          Issue Type: Bug
>          Components: parser
>    Affects Versions: 1.0
>            Reporter: Uwe Schindler
>            Assignee: Chris A. Mattmann
> Lucene/Solr developers ran this tool before releasing Lucene/Solr 3.6 (Solr 3.6 is still
required to run on Java 1.5, see SOLR-3295): http://code.google.com/p/versioncheck/
> {noformat}
> Major.Minor Version : 50.0             JAVA compatibility : Java 1.6 platform: 45.3-50.0
> Number of classes : 60
> Classes are: 
> c:\Work\lucene-solr\.\solr\contrib\extraction\lib\netcdf-4.2-min.jar [:] ucar/unidata/geoloc/Bearing.class
> ...
> {noformat}
> TIKA should use a 1.5 version of this class and especially do some Java 5 tests before
releasing (as it's build dependencies says, it's minimum Java5). I tried to compile and run
TIKA tests with Java 1.5 -> crash (Invalid class file format).

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators: https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira


View raw message