[
https://issues.apache.org/jira/browse/SYSTEMML-1222?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15850636#comment-15850636
]
Deron Eriksson commented on SYSTEMML-1222:
------------------------------------------
Since Long is not a recognized SystemML value type, it makes sense to do the
long to int conversion high up in the API.
Value types in SystemML (from Statement.java):
{code}
public static final String DOUBLE_VALUE_TYPE = "double";
public static final String BOOLEAN_VALUE_TYPE = "boolean";
public static final String INT_VALUE_TYPE = "int";
public static final String STRING_VALUE_TYPE = "string";
{code}
Thank for finding this [~mwdus...@us.ibm.com].
https://github.com/apache/incubator-systemml/pull/370 addresses this.
> MLContext `read` Statement Value Input Error For `cols_in_block` Argument
> -------------------------------------------------------------------------
>
> Key: SYSTEMML-1222
> URL: https://issues.apache.org/jira/browse/SYSTEMML-1222
> Project: SystemML
> Issue Type: Bug
> Reporter: Mike Dusenberry
> Assignee: Deron Eriksson
> Priority: Minor
>
> The {{read}} statement has optional numeric parameters, such as
> {{cols_in_block}} that expect integers. If a Scala {{Long}} is supplied as
> an input variable for one of those parameters via MLContext, and type error
> will occur in which an {{Int}} was expected, but a {{Long}} was received. A
> quick fix is to convert to an integer with {{Long.toInt}} before passing the
> value into the script via MLContext.
> We should automatically cast {{Long}} to {{Int}} internally.
--
This message was sent by Atlassian JIRA
(v6.3.15#6346)
