[
https://issues.apache.org/jira/browse/HADOOP-3785?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12614580#action_12614580
]
Doug Cutting commented on HADOOP-3785:
--------------------------------------
> How about we change scheme and authority to lower cases in the Key
> constructor?
That should work. We create a new Key for nearly every file we touch, but
adding a few string allocations is probably insignificant. There is
String.CASE_INSENSITIVE_ORDER, but I don't know of a case-insensitive hashCode
implementation for strings...
> FileSystem cache should be case-insensitive
> -------------------------------------------
>
> Key: HADOOP-3785
> URL: https://issues.apache.org/jira/browse/HADOOP-3785
> Project: Hadoop Core
> Issue Type: Bug
> Reporter: Doug Cutting
> Fix For: 0.19.0
>
>
> We cache FileSystem instances based on URI scheme & authority, plus username.
> Elsewhere we compare URI schemes and authorities case-insensitively, but the
> cache is case-sensitive. In particular, the Cache.Key#equals() and
> Cache.Key#hashCode() should be made case-insensitive.
> This should not be a blocker, since the worst it causes is a few extra
> FileSystem instances to be cached.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
