You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@systemml.apache.org by "Deron Eriksson (JIRA)" <ji...@apache.org> on 2016/08/12 17:37:20 UTC

[jira] [Assigned] (SYSTEMML-474) Read matrix from URL

     [ https://issues.apache.org/jira/browse/SYSTEMML-474?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Deron Eriksson reassigned SYSTEMML-474:
---------------------------------------

    Assignee: Deron Eriksson

> Read matrix from URL
> --------------------
>
>                 Key: SYSTEMML-474
>                 URL: https://issues.apache.org/jira/browse/SYSTEMML-474
>             Project: SystemML
>          Issue Type: Wish
>          Components: Runtime
>            Reporter: Deron Eriksson
>            Assignee: Deron Eriksson
>
> The ability to read a matrix from a URL can increase usability of SystemML by eliminating the need to have already existing local or cluster data when invoking SystemML.
> For instance, rather than doing something like:
> {code}
> wget http://example.com/ml/matrix.csv
> ./runStandaloneSystemML.sh example.dml -nvargs X=matrix.csv
> {code}
> Someone could do:
> {code}
> ./runStandaloneSystemML.sh example.dml -nvargs X=http://example.com/ml/matrix.csv
> {code}
> One way that this can be useful is to have a small set of example data for each algorithm sitting on a server. Documentation could show a single command that can be invoked for each algorithm that references the server-based example data. This would mean that a user can essentially copy/paste a single command from the documentation, and it would run using the server example data.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)