- From: Benjamin Carlyle <benjamincarlyle@optusnet.com.au>
- Date: Wed, 03 Jan 2007 20:20:47 +1000
- To: Mike Schinkel <mikeschinkel@gmail.com>
- Cc: uri@w3.org
On Tue, 2007-01-02 at 11:48 -0500, Mike Schinkel wrote:
> Benjamin Carlyle wrote:
> > In general, I think that parsing will happen on
> > actual urls rather than url templates.
> Can you give an example? I'm not clear on your thoughts.
Well... just that urls are going to be parsed. We know that. Templates
are going to be substituted. That, too, we know. I'm not sure we know
whether it will be an important use case to actually parse (as opposed
to substituting) a url template. I'm not sure that breaking up a url
template into components is going to be an important feature of the
templating scheme.
For example, we could plausibly break <http://example.com{path}?{query}>
into {http, example.com, {path}, {query}, null} components if we assume
that substitutions cannot occur over component boundaries. How useful
that in general is I think is up for debate. We will certainly need to
subtitute the variable expansions in the unparsed template. Someone
along the way is definitely going to parse the expanded
<http://example.com/the/path?the=query>, too.
I'm not sure that the parsing feature of a uri template is important
enough to limit design choices. I do think that the template should be
limited to the set permitted in a uri[1]. For example, the variable
names could be limited to the pchar set, plus "/" and maybe "?".
Benjamin
[1] or iri. I'm not someone who has come to terms with iris quite yet.
Received on Wednesday, 3 January 2007 10:21:00 UTC