[
https://issues.apache.org/jira/browse/PHOENIX-902?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13950474#comment-13950474
]
Gabriel Reid commented on PHOENIX-902:
--------------------------------------
I noticed the following comment in the patch:
{quote}the primary key columns may still collide with bare columns from column
families.{quote}
Wouldn't it be better to stop if this situation is encountered? If there is a
collision between a primary key and a bare column name that isn't in the key,
the primary key column won't get imported, which (I think) will lead to other
errors a bit further down the line. I think it would be better to fail fast if
a naming collision occurs.
Also, looks like there are tabs for indentation in the patch instead of spaces
-- I think it's supposed to be spaces everywhere.
> Allow family qualified names in CSVCommonsLoader
> ------------------------------------------------
>
> Key: PHOENIX-902
> URL: https://issues.apache.org/jira/browse/PHOENIX-902
> Project: Phoenix
> Issue Type: Bug
> Affects Versions: 3.0.0
> Reporter: James Violette
> Fix For: 3.0.0
>
> Attachments: PHOENIX_902.patch
>
>
> When using family-qualified names, the CSVCommonsLoader cannot find the
> columns.
> We can update the loader to create a family-qualified name map in addition to
> the column name map. When using table-defined columns, the loader would use
> the family-qualified name map. When using user-supplied columns, the loader
> should first look into the family-qualified name map, then in the column name
> map.
--
This message was sent by Atlassian JIRA
(v6.2#6252)
