W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2009

Re: NULL paths in requests

From: Roy T. Fielding <fielding@gbiv.com>
Date: Sun, 5 Jul 2009 14:36:54 -0700
Message-Id: <73D1A10E-E784-405A-9557-316046EF45C1@gbiv.com>
Cc: HTTP Working Group <ietf-http-wg@w3.org>
To: Adrien de Croy <adrien@qbik.com>
On Jul 4, 2009, at 5:07 PM, Adrien de Croy wrote:
>>> Actually absolute-URI as per RFC 3986 seems to allow a lot more  
>>> forms than it did in RFC2616.  Is this desired?  I'd presume not.
>>>
>>> from RFC 3986
>>>
>>>   absolute-URI  = scheme ":" hier-part [ "?" query ]
>>>
>>>   hier-part     = "//" authority path-abempty
>>>                 / path-absolute
>>>                 / path-rootless
>>>                 / path-empty
>>>
>>>
>>> e.g. in the old form, the path-absolute, path-rootless, and path- 
>>> empty forms were not allowed as part of absolute-URI.
>>
>> Please read the ABNF from left to right.
> you mean I should read it as
>
> hier-part = "//" authority (path-abempty / path-absolute / path- 
> rootless / path-empty)
>
> rather than
>
> hier-part = ( "//" authority path-abempty) / (path-absolute) /  
> (path-rootless) / (path-empty)
>
> ?

No, I mean you should read it as

   absolute-URI  = scheme ":" hier-part [ "?" query ]

which clearly shows the scheme and colon before hier-part.
hier-part [ "?" query ] is syntactically equivalent to *uric, and
thus matches the same set of strings as defined by 2616+2396.

....Roy
Received on Sunday, 5 July 2009 21:37:22 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:07 GMT