I see two problems.
If two nodes had the same text attribute value, then you couldn't identify all the nodes.
Second, I think spaces are invalid in URIs.
What about...
http://hosting.opml.org/dave/sta...t=New%
20England
...and if two nodes are the same, then allow an indexer...
http://hosting.opml.org/dave/sta...t=New%
20England[2]
Also, with this syntax, there's no need to drill down.
Thanks for the feedback, Randy. I'll make it so that you can use '+' as well as %20. I'm also planning to add a non-path functionality - so that instead of using a path like "United States/New England", you could just use "New England" - and specify which one you want from a list. The search would only be for nodes with child nodes.
I'll deal with duplicate nodes in much the same way that XPath does. In fact, I'll write it in to the new version that the [x] idea (which is how XPath picks from result sets) is supported.
The reason I put this out was because I had some half-working code sitting there - and because having to constantly split outlines up in to smaller and smaller files is a nuisance for me. I'm going to be building a user interface script as well to make producing OPath scripts easier. All this will arrive shortly, once I've knocked a few real world tasks on the head.
2006-11-12EST00:16:38+00:00
#
I'll deal with duplicate nodes in much the same way that XPath does. In fact, I'll write it in to the new version that the [x] idea (which is how XPath picks from result sets) is supported.
The reason I put this out was because I had some half-working code sitting there - and because having to constantly split outlines up in to smaller and smaller files is a nuisance for me. I'm going to be building a user interface script as well to make producing OPath scripts easier. All this will arrive shortly, once I've knocked a few real world tasks on the head.
I'll need to steal some of your ideas, since coincidentally I've just started playing with something similar:
http://decafbad.com/blog/2006/11...line- addressing
2006-11-13EST19:54:09+00:00
#
http://decafbad.com/blog/2006/11...line- addressing
Why couldn't you just use a fragment id?
So you would have
states.opml#New%20England
This is already in the URI RFC. The interpretation of the the fragment can be different depending on the (MIME) type of the representation, so don't fall in to the trap of thinking that its only for jumps in HTML pages.
I reckon it would be possible to write client-side Javascript that interprets the fragment identifier to give you the chunk of the outline you want.
2006-11-18EST20:54:34+00:00
#
So you would have
states.opml#New%20England
This is already in the URI RFC. The interpretation of the the fragment can be different depending on the (MIME) type of the representation, so don't fall in to the trap of thinking that its only for jumps in HTML pages.
I reckon it would be possible to write client-side Javascript that interprets the fragment identifier to give you the chunk of the outline you want.
Client-side JavaScript wouldn't do the job. How exactly is Grazr or the OPML Editor going to use client-side JavaScript?
As I've said numerous times - this is a tool to popularise a concept. I don't write standards. If people like this idea, I really WANT them to build something better.
2006-11-19EST13:48:00+00:00
#
As I've said numerous times - this is a tool to popularise a concept. I don't write standards. If people like this idea, I really WANT them to build something better.
