Mailing-List: contact dev-help@mahout.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@mahout.apache.org
Message-ID: <3011768.67511293681825752.JavaMail.jira@thor>
Date: Wed, 29 Dec 2010 23:03:45 -0500 (EST)
From: "Hudson (JIRA)" <jira@apache.org>
To: dev@mahout.apache.org
Subject: [jira] Commented: (MAHOUT-545) GSON not a good fit for
 deserializing SGD models
In-Reply-To: <9197360.84031289835134922.JavaMail.jira@thor>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/MAHOUT-545?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12975962#action_12975962 ] 

Hudson commented on MAHOUT-545:
-------------------------------

Integrated in Mahout-Quality #528 (See [https://hudson.apache.org/hudson/job/Mahout-Quality/528/])
    MAHOUT-545 - Big patch to support writable classifiers.


> GSON not a good fit for deserializing SGD models
> ------------------------------------------------
>
>                 Key: MAHOUT-545
>                 URL: https://issues.apache.org/jira/browse/MAHOUT-545
>             Project: Mahout
>          Issue Type: Bug
>          Components: Classification
>    Affects Versions: 0.4
>            Reporter: Ted Dunning
>             Fix For: 0.5
>
>
> The problem is that GSON uses JavaCC which does LL(k) parsing.  The way that the JSON grammar is expressed causes stack recursion for each element of
> a vector.
> This is evil for parsing large models.
> I will be changing ModelSerializer to fix this and will provide a reverse compatible mode for anybody who has been able to serialize/deserialize these models in the past.  My thought is to just make all the SGD models Writables.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.