[
https://issues.apache.org/jira/browse/HADOOP-7531?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13083129#comment-13083129
]
Daryn Sharp commented on HADOOP-7531:
-------------------------------------
bq. Btw that still requires catching exceptions, unlike Jetty's URIUtil all
commons URIUtil encode/decode methods throw URIException, which is lame since
most people want standard byte-by-byte encoding.
Please elaborate because I'm not sure I follow your reasoning... Why would
byte-for-byte be desirable? Perhaps I'm misunderstanding, but if the client
uses a different local charset to encode special chars, then server-side
decoding will fail. Why should this preclude the exception?
Assuming the exception is masked, are you going to change the callers to check
for null and throw an exception? Putting a literal "null" in a URI when
there's an encoding problem is not acceptable. That's messy to handle in the
caller, which is why I think the exception shouldn't be masked.
bq.
Preconditions.checkArgument(request.getRequestURI().startsWith(servletName)).
Otherwise the substring might chop something else off if the function is
misused.
Should this be {{startsWith(servletName+"/")}} to avoid accidently matching a
servlet prefix? Ie. something like "getItem" vs "getItem(s)".
> Add servlet util methods for handling paths in requests
> --------------------------------------------------------
>
> Key: HADOOP-7531
> URL: https://issues.apache.org/jira/browse/HADOOP-7531
> Project: Hadoop Common
> Issue Type: Improvement
> Components: util
> Affects Versions: 0.23.0
> Reporter: Eli Collins
> Assignee: Eli Collins
> Fix For: 0.23.0
>
> Attachments: hadoop-7531-1.patch, hadoop-7531-1.patch,
> hadoop-7531-2.patch, hadoop-7531-3.patch, hadoop-7531-4.patch
>
>
> Common side of HDFS-2235.
--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira
