[
https://issues.apache.org/jira/browse/HADOOP-14785?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16674025#comment-16674025
]
Steve Loughran commented on HADOOP-14785:
-----------------------------------------
Looking at the code of {{FSDataOutputStreamBuilder}}, if you call opt() on a
value after setting must() it gets downgraded to optional, so the order of
setting matters.
I'm not so sure about that: I'd take the view that if something is must(k, v)
then that should stay as mandatory. But we can keep the ordering of settings
sequentials, e,.g
{code}
must("fs:canUnbuffer", true).
opt(("fs:canUnbuffer", false).
{code}
leads to: must("fs:canUnbuffer", false).
Is that good? Or is it confusing?
> Specify the behavior of handling conflicts between must and opt parameters
> ---------------------------------------------------------------------------
>
> Key: HADOOP-14785
> URL: https://issues.apache.org/jira/browse/HADOOP-14785
> Project: Hadoop Common
> Issue Type: Sub-task
> Components: fs
> Affects Versions: 3.0.0-alpha3
> Reporter: Lei (Eddy) Xu
> Priority: Major
>
> It is flexible to allow user to use strings as key/values to specify the
> behaviors of {{FSOutputStream}}, but this flexibility offers the potential
> conflicts between parameters.
> It should specify a general rule of how to handle such conflicts for differnt
> file system implementations.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
