[
https://issues.apache.org/jira/browse/MAHOUT-1310?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13983073#comment-13983073
]
Sean Owen commented on MAHOUT-1310:
-----------------------------------
I also meant that instead of
{code}
File tmpFile = new File(System.getProperty("java.io.tmpdir"));
ModelSerializer.writeBinary(tmpFile.getAbsoluteFile().toString()+System.getProperty("file.separator")+"news-group.model",
{code}
you can do
{code}
ModelSerializer.writeBinary(new File(SystemUtils.getJavaIoTmpDir(),
"news-group.model").toString(),
{code}
but this is pretty trivial.
For the spark module, does it really need its own separate Hadoop dependencies
and different Hadoop 2 profile? I'm not questioning it, just curious.
This is a different question, but does CDH really belong here as a separate
profile, instead of just having a Hadoop 2 profile -- can they be 'merged'? It
shouldn't be a special case.
> Mahout support windows
> ----------------------
>
> Key: MAHOUT-1310
> URL: https://issues.apache.org/jira/browse/MAHOUT-1310
> Project: Mahout
> Issue Type: Task
> Affects Versions: 0.9
> Environment: Operation system: Windows
> Reporter: Sergey Svinarchuk
> Assignee: Sergey Svinarchuk
> Labels: patch
> Fix For: 1.0
>
> Attachments: 1310-2.patch, 1310-3.patch, 1310.patch,
> Mahout_build_on_winodws.PNG
>
>
> Update mahout for build it on Windows with hadoop 2 and add bin/mahout.cmd
--
This message was sent by Atlassian JIRA
(v6.2#6252)
