[
https://issues.apache.org/jira/browse/HIVE-10593?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14550562#comment-14550562
]
Lenni Kuff commented on HIVE-10593:
-----------------------------------
[~singhashish] / [~rdblue] - Agree that we should avoid a table property
because it doesn't seem necessary. Allowing a directory to be passed in is a
bit more complicated than a file because you might have multiple files with
different schemas in the directory and you need to add logic to verify the
schemas match. I would suggest initially supporting file-only.
Do you have feedback on the syntax described in this JIRA? It seems like it
avoids the use of table properties and the syntax is fairly natural (similar to
CREATE TABLE LIKE otherTable).
> Support creating table from a file schema: CREATE TABLE ... LIKE
> <file_format> '/path/to/file'
> ----------------------------------------------------------------------------------------------
>
> Key: HIVE-10593
> URL: https://issues.apache.org/jira/browse/HIVE-10593
> Project: Hive
> Issue Type: Improvement
> Components: Query Processor
> Affects Versions: 1.2.0
> Reporter: Lenni Kuff
>
> It would be useful if Hive could infer the column definitions in a create
> table statement from the underlying data file. For example:
> CREATE TABLE new_tbl LIKE PARQUET '/path/to/file.parquet';
> If the targeted file is not the specified file format, the statement should
> fail analysis. In addition to PARQUET, it would be useful to support other
> formats such as AVRO, JSON, and ORC.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
