In our last episode,
<ef504f7e-5e47-48a1-9540->, the
lovely and talented
broadcast on alt.html:
> Hi,
> While parsing relative URL segments "../" to hierarchical segments of
> the absolute URL , I notice that the <a> tag and the <link> tag do
> this differently. Take the following absolute and relative URLs:
> absolute URL: http://www.a.com/a/b
> relative URL: ../../b/c.css
But this is nonsense. It is looking for the parent of the root,
but of course there cannot be one.
> The <a> tag resolves this to:
> http://www.a.com/b/c.css
> However the <link> tag resolves this to:
> http://www.a.com/a/b/c.css
In other words, some browser or another, presented with an absurdity, has
inconsistent error handling. Well, obviously, other than writing to the
authors of the browser about that, the solution would be not to write
relative URLs that are absurd.
> It appears that all other forms of resolving URLs are equal between
> both tags however, I have been trying to find the RFC specification on
> this but have had no luck. This document http://www.ietf.org/rfc/rfc2396.txt
> does not contain any information regarding the <link> tag.
I don't think you will find many specs that deal with error recovery. There
are so many possible errors that the point of defining the right way would
be lost in all the stuff about what to do when people do things the wrong
way.
> Is anyone here familiar with this or know where I could obtain more
> information.
-
Lars Eighner <http://larseighner.com/>
Countdown: 232 days to go.