You are viewing a plain text version of this content. The canonical link for it is here.

Posted to dev@metamodel.apache.org by "Kasper Sørensen (JIRA)" <ji...@apache.org> on 2014/09/29 21:06:34 UTC

[jira] [Commented] (METAMODEL-82) Metadata type is always VARCHAR for CSV & Excel modules

    [ https://issues.apache.org/jira/browse/METAMODEL-82?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14152082#comment-14152082 ] 

Kasper Sørensen commented on METAMODEL-82:
------------------------------------------

Well yes and no. It's a tough one...

The trouble is that the data here can never be guaranteed as any other type. If you put a number in a CSV file or an Excel sheet, it can only be treated as a number if all the other rows have a number in this column. And how do we guarantee that (except for obviously eager-reading all records, which isn't acceptable IMO).

If we change this I would suggest to make a "type detection" optional (and default disabled for compatibility). Then we could eager-read up to 1000 records or so and use the schema inference utils to determine a pretty precise type.

> Metadata type is always VARCHAR for CSV & Excel modules
> -------------------------------------------------------
>
>                 Key: METAMODEL-82
>                 URL: https://issues.apache.org/jira/browse/METAMODEL-82
>             Project: Metamodel
>          Issue Type: Bug
>            Reporter: Alberto Rodriguez
>
> The metadata types fetched from CSV and Excel DataContexts are always VARCHAR although the data of some columns are not a string.



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