- From: Robin Berjon <robin@berjon.com>
- Date: Mon, 6 Jul 2009 15:53:51 +0200
- To: Marcos Caceres <marcosc@opera.com>
- Cc: public-webapps WG <public-webapps@w3.org>
On Jul 1, 2009, at 10:34 , Marcos Caceres wrote:
> On 6/30/09 7:49 PM, timeless wrote:
>> speaking of which,
>>
>> does<a href="foo/">foo</a> magically map to a file?
>>
>> i'm assuming we're going to wait until the widget uri scheme suggests
>> a handling for that.
>
> Yes, I think so.
I would tend to think that it should map to nothing (i.e. it's an
error, or a faulty reference, or whatever it is we call these things).
The justification for this position is that there are only three ways
in which to handle a reference to a directory (either as href or src):
A. Have a mapping to a default file. This would mean we'd have to
specify a list of filenames that would match in order (e.g.
index.html, index.xhtml, index.html, index.svg, index.svgz...). That
list would be arbitrary (i.e. it would fail the test of independent
invention) and likely the source of a healthy bikeshed discussion.
What's more, each file name would have to be tested not just for the
given directory but also for its localised alternatives. That's just
extra complexity and de facto poor spec work when the author can just
pull the finger out and type the extra few characters
B. List the directory's content. For href links that's reasonably easy
as the UA can just display the list whichever way it wants, as some
already do for file:. That being said, we probably don't want end
users to navigate the widget's content. For src links it gets worse:
you need a format to represent directories. Not hard, but we surely
don't want to go there.
C. Ignore the request and do nothing. Since the only two other options
don't work, this one is simply the best, better than all the rest.
--
Robin Berjon - http://berjon.com/
Feel like hiring me? Go to http://robineko.com/
Received on Monday, 6 July 2009 13:54:34 UTC