Re: up up up, was: Last Call: draft-nottingham-http-link-header (Web Linking) to Proposed Standard

Ian Hickson On 09-08-31 12.18:

> On Mon, 31 Aug 2009, Julian Reschke wrote:
>> Ian Hickson wrote:
>>>> On the other hand, the advantage you are citing is only an advantage 
>>>> if a given resource only contains "up up up", but not also "up up" 
>>>> and "up". What you be the point of that?
>>> A UA that treats any number of "up"s as the same would not interpret 
>>> "up" as "up1", but as meaning that the referenced document was some 
>>> higher- level document in the hierarchy.
>> Again, what would be the point of this case ("up up up" being present, 
>> but not "up up" and "up")?
> 
> Why would they not be present?

It is striking that only the use-case that you have problems 
comprehending gets benefit from the "up up up" design: Only in 
that particular case would user agents that look for just "up" as 
well as user agents that also look for "up up up", interpret 
rel="up up up" correctly - judged from their own perspective, and 
with certainty - judged from users point of view.

Ian Hickson On 09-08-31 08.23:

> A UA that treats any number of "up"s as the same would not interpret "up" 
> as "up1", but as meaning that the referenced document was some higher- 
> level document in the hierarchy.


"Up" is currently used as "_the_ up", which is semantically 
different from "something up". Additionally, rel="up" is most 
often taken to mean "one level up". (As long as there is only one 
"up", it becomes less important whether this "one level up" equals 
exactly one directory level up or not.)

A possible workaround could be to recommend that when a single 
"up" and multiple "up"s appear in the same context, then the 
single "up" should appear last in source code. (But there will of 
course be cases when this doesn't work - e.g. some might want to 
have the breadcrumb go from right to left.)

Ian Hickson on 30th of August or before (in some other list):

>>> Also, it means that we don't have to register an infinite
>> > number of keywords for all possible depths.

Wouldn't there be other ways to create a syntax that doesn't need 
to register an infinite number of keywords? E.g. just design a 
bind character to place between each "up". For example

rel="up+up+up"

or

rel="up>up>up"

(At least iCab will interpret either of those as just "up". In 
fact, iCab will even interpret "up3" as "up" ... )
-- 
leif halvard silli

Received on Monday, 31 August 2009 19:54:33 UTC