mahout-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Sean Owen <sro...@gmail.com>
Subject Re: [jira] [Commented] (MAHOUT-1396) Accidental use of commons-math won't work with next Hadoop 2 release
Date Wed, 15 Jan 2014 08:34:21 GMT
I think you can pause a bit to see if anything else comes up then create
another build.

I think you can drop the staging repo and use release:rollback and then
start again? As I say it I am not sure I have tried rolling back after
release:perform. I wonder if the right process is to release artifacts with
names including RCn and then at the last make the final release. Then no
rollback.
On Jan 15, 2014 8:09 AM, "Suneel Marthi (JIRA)" <jira@apache.org> wrote:

>
>     [
> https://issues.apache.org/jira/browse/MAHOUT-1396?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13871782#comment-13871782]
>
> Suneel Marthi commented on MAHOUT-1396:
> ---------------------------------------
>
> Sean, I think we should be able to get this into 0.9 release. Since we
> have a candidate 0.9 Release that's presently in staging do we need to roll
> that back before any code can be committed for 0.9 into SVN?  I think we
> should get this and M-1394 into 0.9.
>
>
>
> > Accidental use of commons-math won't work with next Hadoop 2 release
> > --------------------------------------------------------------------
> >
> >                 Key: MAHOUT-1396
> >                 URL: https://issues.apache.org/jira/browse/MAHOUT-1396
> >             Project: Mahout
> >          Issue Type: Bug
> >          Components: Classification
> >    Affects Versions: 0.8
> >            Reporter: Sean Owen
> >         Attachments: MAHOUT-1396.patch
> >
> >
> > The project uses commons-math3, since about a year ago. However there is
> a use of old commons-math (2.2) lurking:
> > core/src/main/java/org/apache/mahout/classifier/sgd/TPrior.java:
> > import org.apache.commons.math.special.Gamma;
> > This happens to have worked since commons-math has been pulled in by
> hadoop-common. But it no longer is in HEAD:
> >
> http://svn.apache.org/viewvc/hadoop/common/trunk/hadoop-common-project/hadoop-common/pom.xml?view=markup
> > So this will no longer compile against the latest Hadoop. I believe it
> will also not actually run again the latest Hadoop, even if one were to use
> a version compiled versus older Hadoop 2, since the class that uses it is
> used in the context of Writables -- that is, outside the client environment
> that might happen to have packaged commons-math -- and so would fail on the
> cluster.
> > The change is trivial, to import the commons-math3 class. I've verified
> that tests pass and a patch is attached.
> > Question is how much of a 'blocker' this should be for the pending
> release. It would cause it to stop working with the next Hadoop 2 release,
> so would be useful to get in, IMHO.
>
>
>
> --
> This message was sent by Atlassian JIRA
> (v6.1.5#6160)
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message