You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@systemml.apache.org by "Matthias Boehm (JIRA)" <ji...@apache.org> on 2017/06/07 05:06:18 UTC

[jira] [Commented] (SYSTEMML-947) Simplification new mlcontext external data structures

    [ https://issues.apache.org/jira/browse/SYSTEMML-947?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16040185#comment-16040185 ] 

Matthias Boehm commented on SYSTEMML-947:
-----------------------------------------

thanks for taking this over [~deron] - channeling all potential inputs through one abstract type would also allow us to make various data conversions lazy. This will help in avoiding useless conversions we currently see at API level such as csv->dataframe->binary block (distributed)-> binary block (local), which should boil down to csv -> binary block (local).

> Simplification new mlcontext external data structures
> -----------------------------------------------------
>
>                 Key: SYSTEMML-947
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-947
>             Project: SystemML
>          Issue Type: Task
>            Reporter: Matthias Boehm
>            Assignee: Deron Eriksson
>
> This task aims to simplify the externally exposed data structures of the new MLContext API. The details are discussed here: https://www.mail-archive.com/dev@systemml.incubator.apache.org/msg00795.html. The goal is to remove MatrixObject/FrameObject from the API, remove BinaryBlockMatrix/BinaryBlockFrame, and provide all functionalities through the abstract types Matrix and Frame.



--
This message was sent by Atlassian JIRA
(v6.3.15#6346)