On 10/22/05, Robert Sayre <[EMAIL PROTECTED]> wrote:
> On 10/22/05, Luke Arno <[EMAIL PROTECTED]> wrote:
> > I still think control should be an extension.
> >
> > http://www.intertwingly.net/wiki/pie/PaceDropControl
> >
> > These are the two proposals I will make regarding
> > basic-04:
> >
> > 1. Use rel edit rather then an edit element.
>
> Ok, but let me try to sell you. Below, there's a simple diagram. Think
> of the source element as whatever you would send PUT and DELETE to.
> When I see an entry in an APP feed, I like to think that it's telling
> me there's a source resource that's probably turned into at least one
> or two public/derived resources by some Turing-complete process that
> the diagram calls the transform.
>
>                        -> derived
>                       /
>                      /
> source --> transform ---> derived
>                      \
>                       \
>                        -> derived
>
> The entry in that APP feed is telling you that there's a source.

Sure. Of course.

> I think atom:link
> is good for pointing downstream or sideways (from derived to derived),
> but not so much for pointing upstream.

Why? <scratches head/>

> In a Movable Type installation,
> the source resources live behind a CGI, the transform is the
> template+perl, and the derived resources are the static things that
> get spit out by the build process. So, you get something like this
> (looks better in monospace):
>
>     APP feed
>        |
>        |               -> derived
>        V              /
>        |             /
> source --> transform ---> derived
>        |             \
>        |              \
>        V               -> derived
>        |
>        |               -> derived
>        V              /
>        |             /
> source --> transform ---> derived
>        |             \
>        |              \
>        V               -> derived
>        |
>        |               -> derived
>        V              /
>        |             /
> source --> transform ---> derived
>        |             \
>        |              \
>        V               -> derived
>        ...
>

I know all this.

I don't see what that tells me about link vs. edit.

What am I missing?

> > 2. Get rid of control.
>
> Well, if you give people a specific place to put their extension, they
> can say they are using the standard exactly as designed. It seems like
> kind of a lame security blanket to me, but maybe it will be really
> important. If it turns out to be a bad idea, we can drop it from the
> protocol (the IETF process is designed to handle this).
>

I still think it should be an extension.

> > If we can get away with not adding anything to
> > atom documents to make the protocol work, that
> > would be beautiful.
>
> I see that as a non-goal.
>

It is not a goal but it smells nice. I trust my nose.

Your aesthetic may vary.

> > That said, I prefer basic by far.
>
> Most people do.
>

Good.

- Luke

Reply via email to