You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@river.apache.org by Mark Brouwer <ma...@cheiron.org> on 2008/01/29 21:37:53 UTC

Re: [jira] Commented: (RIVER-5) HTTMPMD URLs can be considered equal while they are not

Thanks Fred for the review,

Fred Oliver (JIRA) wrote:
>     [ https://issues.apache.org/jira/browse/RIVER-5?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12563336#action_12563336 ] 
> 
> Fred Oliver commented on RIVER-5:
> ---------------------------------
> 
> First, the parseURL() method has the same bug. I would prefer that method be
> fixed as well.

I don't see the bug in parseURL, there is a test for the mandatory '=' 
character that separates the digest algorithm from the digest value. The 
comment is looked for relative from the '=' character.

> Second, the fix in the patch treats the '=' characters as optional instead of
> required. I would prefer that the method explicitly return false if the '='
> characters are missing from either string, or not located at the same index.

You are right, will do that.
-- 
Mark

Re: [jira] Commented: (RIVER-5) HTTMPMD URLs can be considered equal while they are not

Posted by Mark Brouwer <ma...@cheiron.org>.
Mark Brouwer wrote:

> I don't see the bug in parseURL, there is a test for the mandatory '=' 
> character that separates the digest algorithm from the digest value. The 
> comment is looked for relative from the '=' character.

By looking more careful at the java.net.URL implementation I now
understand the code block executed in parseURL for "a relative URL that
only specifies a comment".

>> Second, the fix in the patch treats the '=' characters as optional 
>> instead of
>> required. I would prefer that the method explicitly return false if 
>> the '='
>> characters are missing from either string, or not located at the same 
>> index.
> 
> You are right, will do that.

I also modified Freds test class so it's easier to run and doesn't rely
on a Sun specific security provider. It begs the question what we are
going to do with these unit tests ...

It looks like Fred and I agree on the fix so I will commit it in a few
hours.
-- 
Mark