From: Ian Hickson (ian@hixie.ch)
Date: Mon Sep 09 2002 - 21:04:51 BST
On Mon, 9 Sep 2002, Jim Dabell wrote:
> On Sunday 08 September 2002 10:05 pm, Aquarion wrote:
> > On Sun, Sep 08, 2002 at 06:29:31PM +0000, Jim Dabell wrote:
> >
> > No, I don't. I'm opposed to the idea of multiple methodologies for this
> > anyway, and even more opposed to the idea of having to parse three
> > differant documents in order do do another weblog what is, basically, a
> > favour. Every other protocal you /have/ to impliment in order for a
> > client to work in the Real World makes it less likely that a given
> > developer will bother with it.
>
> The linking html, the xml file... what's the third document?
The three documents are:
The source document (the blog entry being written, the linker).
The target document (the one being pinged back, the linkee).
Your XML file.
> > > > The whole point of the original
> > > > spec was that all the client had to do was send off a single RPC-XML
> > > > request saying it was linked to. This is quick, this is efficent,
> > > > this is /easy/.
> > >
> > > It's also non-extendable. Doing the exact same thing, except grabbing
> > > the xml file to get the location of the rpc file is not that much
> > > different (assuming no namespaces).
> >
> > Er, yes it is. As I mentioned, now I have to grab andparse the headers,
> > possibly the document as well,
>
> Well you get the headers "for free", they are an optional part of the spec
> (so there's no "have to"), and you were parsing the html document anyway!
Actually the headers are a MUST in the current spec.
> > and grab and parse the resulting file
> > until I even know what server I'm connecting to.
>
> This bit is the difference. I don't see it as being particularly onerous,
> but if you do, then you can simply not implement the generic bit.
Well if you don't implement the generic bit there isn't much point having
it, is there...
If you want extensibility, I don't see that the current system is a
blocker. Here are various ways a future version of the spec could extend
pingback, in the unlikely (IMHO) case that that is wanted:
1. Extend the XML-RPC API by adding new methods.
2. Change the name of the link type and the HTTP header.
3. Make the spec allow two headers and two link tags, and tell the
version 2 clients to use the second header/tag.
4. Change the target file to be an XML file instead of an XML-RPC
processor. Clients can try to get the fil using GET, and if that
fails (by a 405, probably) then they should use it as an XML-RPC
server.
5. Invent an algorithm for autodiscovery that works much the same way
as the P3P autodiscovery (i.e. guess at the URI of the XML file)
6. etc.
I don't mind making it extensible. I do mind making the spec hard for blog
authors to use (they have to include an XML file _and_ a link tag?).
-- Ian Hickson )\._.,--....,'``. fL "meow" /, _.. \ _\ ;`._ ,. http://index.hixie.ch/ `._.-(,_..'--(,_..'`-.;.' Message sent over the Blogite mailing list. Archives: http://www.aquarionics.com/misc/archives/blogite/ Instructions: http://www.aquarionics.com/misc/blogite/
This archive was generated by hypermail 2.1.5 : Mon Sep 09 2002 - 22:05:00 BST