arrow-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Wes McKinney (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (ARROW-322) [C++] Do not build HDFS IO interface optionally
Date Mon, 26 Dec 2016 18:29:58 GMT

    [ https://issues.apache.org/jira/browse/ARROW-322?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15778776#comment-15778776
] 

Wes McKinney commented on ARROW-322:
------------------------------------

[~xhochy] since there's no extra thirdparty dependency and these libraries are loaded only
at runtime using dlopen, there's little harm in always building the module. If it becomes
an issue for some reason in the future, we can split it out into a leaf shared library

> [C++] Do not build HDFS IO interface optionally
> -----------------------------------------------
>
>                 Key: ARROW-322
>                 URL: https://issues.apache.org/jira/browse/ARROW-322
>             Project: Apache Arrow
>          Issue Type: Bug
>          Components: Python
>            Reporter: Wes McKinney
>
> Currently if you have not built {{arrow::io}} with {{-DARROW_HDFS=on}} then importing
{{pyarrow.parquet}} fails with a not very graceful error message from failure to find the
right symbols in {{libarrow_io.so}}. Since we vendor {{hdfs.h}} and the libraries are loaded
at runtime, having this extra option doesn't add much value. 



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

Mime
View raw message