- From: Roy T. Fielding <fielding@gbiv.com>
- Date: Fri, 13 Feb 2004 18:06:07 -0800
- To: Graham Klyne <gk@ninebynine.org>
- Cc: uri@w3.org
On Thursday, June 19, 2003, at 02:45 AM, Graham Klyne wrote: > (1) hierarchical and non-hierarchical URIs > > I notice that in -03 the opaque-part syntax distinction has been > dropped. My concern is that it may now not be clear when > relative-to-absolute URI conversion should take part of hierarchical > '/' characters in the URIs. Previously, my understanding was that an > algorithm would look at the path component of the base URI and, if it > starts with a '/', assume the base and relative URIs are hierarchical > and apply the path-merging logic; otherwise, the opaque-part in the > base URI is used in all-or-nothing fashion depending on what is > present in the "relative" URI. I've never used the base URI's path as an indicator. The problem with that theory is folks wanted to eliminate dot-segments from paths regardless of reference type. > Section 3 says "a non-hierarchical path will be treated as opaque data > by a generic URI parser", but it's not clear at this point what > constitutes a "non-hierarchical path". fixed. > Section 3.3 says " A path is always defined for a URI, though the > defined path may be empty (zero length) or opaque (not containing any > "/" delimiters)", which suggests that an opaque path may not contain > *any* un-escaped '/' characters. This seems like an onerous > restriction, and in conflict with existing URI scheme usage; e.g. > news: according to IANA is currently specified by RFC 1738 fixed. > (2) square brackets > > Is it necessary for square brackets to be reserved outside the > net-path component? I personally use them quite often in fragment > identifiers for references. My correspondent had another use for them > in the path component of a URI scheme. I think there are several > instances of them occurring (unescaped) in message-IDs, and the mid: > spec doesn't require them to be escaped. I think this also applies to > the ldap: scheme. > -- http://www.faqs.org/rfcs/rfc2392.html Says they must be percent-encoded (section 2). > -- http://www.faqs.org/rfcs/rfc2255.html Note that any URL-illegal characters (e.g., spaces), URL special characters (as defined in section 2.2 of RFC 1738) and the reserved character '?' (ASCII 63) occurring inside a dn, filter, or other element of an LDAP URL MUST be escaped using the % method described in RFC 1738 [5]. If a comma character ',' occurs inside an extension value, the character MUST also be escaped using the % method. Nope. Square brackets have never been allowed in a URI before, so allowing them in the path may lead to errors in implementations that assumed they could use them internally. ....Roy
Received on Friday, 13 February 2004 21:05:16 UTC