[
https://issues.apache.org/jira/browse/HADOOP-8756?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Colin Patrick McCabe updated HADOOP-8756:
-----------------------------------------
Description:
* We use {{System.loadLibrary("snappy")}} from the Java side. However in
libhadoop, we use {{dlopen}} to open libsnappy.so dynamically. It seems like
the System.loadLibrary nonsense is completely redundant-- we would only need
this if we planned on accessing libsnappy methods directly from Java (via the
'native' keyword)- - which we don't, and can't, since snappy is not JNI. At
the end of the day, snappy is only accessible through libhadoop.so, so I think
snappy-related issues should be handled there.
* We should log the search path(s) we use for {{libsnappy.so}}, so that it's
easier to diagnose configuration issues.
was:
* We use {{System.loadLibrary("snappy")}} from the Java side. However in
libhadoop, we use {{dlopen}} to open libsnappy.so dynamically. It seems like
the System.loadLibrary nonsense is completely redundant-- we would only need
this if we planned on accessing libsnappy methods directly from Java (via the
'native' keyword)- - which we don't, and can't, since snappy is not JNI. At
the end of the day, snappy is only accessible through libhadoop.so, so I think
snappy-related issues should be handled there.
* We should log the search path(s) we use for {{libsnappy.so}}, so that it's
easier to diagnose configuration issues.
* It seems like the static initialization code in the LoadSnappy.java class
references the static initialization code in NativeCodeLoader. Basically, if
the init method in NativeCodeLoader doesn't run before the init method in
LoadSnappy, there could be problems. We should avoid having this kind of issue.
> libsnappy loader issues
> -----------------------
>
> Key: HADOOP-8756
> URL: https://issues.apache.org/jira/browse/HADOOP-8756
> Project: Hadoop Common
> Issue Type: Bug
> Components: native
> Affects Versions: 2.2.0-alpha
> Reporter: Colin Patrick McCabe
> Assignee: Colin Patrick McCabe
> Priority: Minor
>
> * We use {{System.loadLibrary("snappy")}} from the Java side. However in
> libhadoop, we use {{dlopen}} to open libsnappy.so dynamically. It seems like
> the System.loadLibrary nonsense is completely redundant-- we would only need
> this if we planned on accessing libsnappy methods directly from Java (via the
> 'native' keyword)- - which we don't, and can't, since snappy is not JNI. At
> the end of the day, snappy is only accessible through libhadoop.so, so I
> think snappy-related issues should be handled there.
> * We should log the search path(s) we use for {{libsnappy.so}}, so that it's
> easier to diagnose configuration issues.
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira
