[
https://issues.apache.org/jira/browse/BEAM-292?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15317839#comment-15317839
]
ASF GitHub Bot commented on BEAM-292:
-------------------------------------
GitHub user dhalperi opened a pull request:
https://github.com/apache/incubator-beam/pull/424
[BEAM-292] Write: always produce at least 1 WriteT
Write has a degenerate case wherein, if no elements were in the written
PCollection,
the finalize step will get produced with nothing to finalize. This often
prevents
correct operation, for example when a FileBasedSink produces no files
instead of one
empty file.
Catch and handle this case in Write by opening and closing an empty Writer
to
produce a single WriteT.
You can merge this pull request into a Git repository by running:
$ git pull https://github.com/dhalperi/incubator-beam write-numshards
Alternatively you can review and apply these changes as the patch at:
https://github.com/apache/incubator-beam/pull/424.patch
To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:
This closes #424
----
commit 2cd79b6c760ed6aeae41a9b69bc628cd72a4ac2d
Author: Dan Halperin <dhalp...@google.com>
Date: 2016-06-07T04:46:37Z
[BEAM-292] Write: always produce at least 1 WriteT
Write has a degenerate case wherein, if no elements were in the written
PCollection,
the finalize step will get produced with nothing to finalize. This often
prevents
correct operation, for example when a FileBasedSink produces no files
instead of one
empty file.
Fix this in Write by opening and closing an empty Writer to produce a single
Write.
----
> TextIO.Write.to Empty Files
> ---------------------------
>
> Key: BEAM-292
> URL: https://issues.apache.org/jira/browse/BEAM-292
> Project: Beam
> Issue Type: Bug
> Components: sdk-java-core
> Reporter: Jesse Anderson
>
> When a PCollection is empty and is written out with TextIO.Write.to, the
> output file is unchanged. This makes it seem like the PCollection was not
> empty. The output file's contents should be changed to be empty.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
