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 2015/07/16 23:20:05 UTC
[jira] [Created] (METAMODEL-163) Composite/directory Resource for
local files and HDFS files
Kasper Sørensen created METAMODEL-163:
-----------------------------------------
Summary: Composite/directory Resource for local files and HDFS files
Key: METAMODEL-163
URL: https://issues.apache.org/jira/browse/METAMODEL-163
Project: Apache MetaModel
Issue Type: Improvement
Reporter: Kasper Sørensen
A more and more common pattern in representing data is to have a directory with files of the same format which can be appended together to form a complete dataset. I see this especially in Hadoop scenarios where reducers as well as spark usually will create such "part" files in a directory and treat that directory almost as a logical file.
I don't know if we can generalize this or if we need two separate implementations. But at least I would love to have a Resource implementation like this: Given a (local or HDFS) path that points to a directory, or maybe also to a wildcard-enabled expression, I would want to have a single Resource object that represents all the corresponding files in that directory/pattern.
This would not only provide us with better interoperability with Hadoop result data, but it will also actually solve a long-standing request (in our company at least) to support multiple CSV files in one logical CsvDataContext.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)