You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@mahout.apache.org by "Sean Owen (JIRA)" <ji...@apache.org> on 2011/04/10 23:59:05 UTC
[jira] [Created] (MAHOUT-664) Remove usage of XStream string
serialization too?
Remove usage of XStream string serialization too?
-------------------------------------------------
Key: MAHOUT-664
URL: https://issues.apache.org/jira/browse/MAHOUT-664
Project: Mahout
Issue Type: Improvement
Affects Versions: 0.4, 0.5
Reporter: Sean Owen
Priority: Minor
Fix For: 1.0
In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
There are two key mahout-core usages:
- org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
- org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
Some usages are serialization of longs and char[], which seems pointless.
Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Updated] (MAHOUT-664) Remove usage of XStream string
serialization too?
Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAHOUT-664?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sean Owen updated MAHOUT-664:
-----------------------------
Due Date: 30/Jun/11
Fix Version/s: (was: 1.0)
0.6
Assignee: Sean Owen
> Remove usage of XStream string serialization too?
> -------------------------------------------------
>
> Key: MAHOUT-664
> URL: https://issues.apache.org/jira/browse/MAHOUT-664
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.4, 0.5
> Reporter: Sean Owen
> Assignee: Sean Owen
> Priority: Minor
> Fix For: 0.6
>
>
> In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
> There are two key mahout-core usages:
> - org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
> - org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
> All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
> Some usages are serialization of longs and char[], which seems pointless.
> Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAHOUT-664) Remove usage of XStream string
serialization too?
Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAHOUT-664?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13042770#comment-13042770 ]
Sean Owen commented on MAHOUT-664:
----------------------------------
I can't really get any farther on this issue since removing XStream would mean removing serialization of TreeBuilder and FitnessEvaluator, and the way the code is structured it does require it.
We could use Writable, or accept a use of Serializable here, except that the result needs to be a String. that can be base-64 encoded, but then I wonder if it's getting too weird? versus keeping the dependency on XPP and Xstream just for this. I really don't know, any thoughts?
> Remove usage of XStream string serialization too?
> -------------------------------------------------
>
> Key: MAHOUT-664
> URL: https://issues.apache.org/jira/browse/MAHOUT-664
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.4, 0.5
> Reporter: Sean Owen
> Assignee: Sean Owen
> Priority: Minor
> Fix For: 0.6
>
>
> In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
> There are two key mahout-core usages:
> - org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
> - org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
> All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
> Some usages are serialization of longs and char[], which seems pointless.
> Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAHOUT-664) Remove usage of XStream string
serialization too?
Posted by "Hudson (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAHOUT-664?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13020323#comment-13020323 ]
Hudson commented on MAHOUT-664:
-------------------------------
Integrated in Mahout-Quality #751 (See [https://hudson.apache.org/hudson/job/Mahout-Quality/751/])
First step on MAHOUT-664, remove redundant use of Xstream
> Remove usage of XStream string serialization too?
> -------------------------------------------------
>
> Key: MAHOUT-664
> URL: https://issues.apache.org/jira/browse/MAHOUT-664
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.4, 0.5
> Reporter: Sean Owen
> Priority: Minor
> Fix For: 1.0
>
>
> In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
> There are two key mahout-core usages:
> - org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
> - org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
> All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
> Some usages are serialization of longs and char[], which seems pointless.
> Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Commented] (MAHOUT-664) Remove usage of XStream string
serialization too?
Posted by "Deneche A. Hakim (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAHOUT-664?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13037272#comment-13037272 ]
Deneche A. Hakim commented on MAHOUT-664:
-----------------------------------------
I will have to take a closer look to the watchmaker code, but for df I know I used StringUtils to serialize some implementation of TreeBuilder, for now it's just DefaultTreeBuilder but it can be a custom implementation too.
> Remove usage of XStream string serialization too?
> -------------------------------------------------
>
> Key: MAHOUT-664
> URL: https://issues.apache.org/jira/browse/MAHOUT-664
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.4, 0.5
> Reporter: Sean Owen
> Assignee: Sean Owen
> Priority: Minor
> Fix For: 0.6
>
>
> In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
> There are two key mahout-core usages:
> - org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
> - org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
> All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
> Some usages are serialization of longs and char[], which seems pointless.
> Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
[jira] [Resolved] (MAHOUT-664) Remove usage of XStream string
serialization too?
Posted by "Sean Owen (JIRA)" <ji...@apache.org>.
[ https://issues.apache.org/jira/browse/MAHOUT-664?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Sean Owen resolved MAHOUT-664.
------------------------------
Resolution: Fixed
I give up, I don't think there is way to meaningfully finish this, so, having removed the superfluous usages, am calling it a day.
> Remove usage of XStream string serialization too?
> -------------------------------------------------
>
> Key: MAHOUT-664
> URL: https://issues.apache.org/jira/browse/MAHOUT-664
> Project: Mahout
> Issue Type: Improvement
> Affects Versions: 0.4, 0.5
> Reporter: Sean Owen
> Assignee: Sean Owen
> Priority: Minor
> Fix For: 0.6
>
>
> In the spirit of MAHOUT-510 and continuing to standardize approaches without going too nuts -- I notice that there is only one use of XStream left in the code, in StringUtils, which is merely a wrapper for serializing/deserializing an object to a String.
> There are two key mahout-core usages:
> - org.apache.mahout.ga.watchmaker. Here it seems to be used to serialize/deserialize a candidate solution to a file on HDFS and fed into a mapper. Surely it's more appropriate to do something with Writable here?
> - org.apache.mahout.df.mapreduce. It's used to serialize/deserialize a DefaultTreeBuilder. Does it need to be done? it seems like DefaultTreeBuilder is parameterized entirely by its param "m" but I haven't looked hard.
> All of the other usages are in mahout-examples/, which suggests they're of secondary importance.
> Some usages are serialization of longs and char[], which seems pointless.
> Any thoughts? putting this on the map for 1.0.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira