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

Re: [draft-nottingham-http-link-header-06] relation-types

From: Julian Reschke <julian.reschke@gmx.de>
Date: Tue, 25 Aug 2009 10:08:55 +0200
Message-ID: <4A939C17.2040705@gmx.de>
To: Anne van Kesteren <annevk@opera.com>
CC: Mark Nottingham <mnot@mnot.net>, HTTP Working Group <ietf-http-wg@w3.org>
Anne van Kesteren wrote:
> On Fri, 21 Aug 2009 08:34:25 +0200, Mark Nottingham <mnot@mnot.net> wrote:
>> On 25/07/2009, at 12:05 AM, Anne van Kesteren wrote:
>>> I think multiple SP characters should be allowed between relation  
>>> types. That makes it more consistent with other serializations.
>>> I actually doubt existing implementations do it like the draft  
>>> suggests. The tab character would probably work fine too and maybe  
>>> something else I'm forgetting. Has this been tested?
>> Existing implementations do it like the draft suggests, in that when  
>> they use the Link header, they accept the specified form.
>>
>> FF3.5 allows both multiple spaces and tabs.
>> Opera 10b3 only allows multiple spaces, doesn't recognise tabs.
>> Safari and IE both appear to ignore the Link header.
>>
>> The safest thing to do is to leave it specified as a single space; all  
>> implementations that support Link (remembering that it was first defined  
>> this way in 2068) will be able to consume that.
> 
> Well, I do not think that is the safest thing at all from my perspective:
> 
> 1. It is not clear to me what it means to recognize only a single space as consumer. What are the values for "foo{SP}{SP}bar"? Will that be "foo" and "{SP}bar"? Or "foo{SP}" and "bar"? Something else?
> ...

It will be invalid.

 > ...

BR, Julian
Received on Tuesday, 25 August 2009 08:09:37 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:08 GMT