[uf-new] Comments proposals - the issue of rel="in-reply-to"

[David Janes]

I'm splitting Sarven's brain dump into multiple responses (with
difference subject lines)

On Sun, Nov 16, 2008 at 5:56 PM, Sarven Capadisli <csarven at gmail.com> wrote:
> Okay, so, here is my brain dump:
>
> David makes a valid point; we should put more focus back on the
> existing examples in the wild.
>
> I would like to get more insight from you (community) on why grabbing
> the URI out of rel="in-reply-to" is semantically incorrect as David
> mentioned [1]. Isn't this the case for all @rel values?

There are two cases here that have to be considered independently.

In both cases, consider a blog post A identified by URL-A, and a
comment B (on A) identified by URL-B.

= Case 1 =

We add to B:
<a rel="in-reply-to" href="URL-A">my parent</a>

This semantically correct, but 0% of the examples given on the Wiki
have elements that you can add this to (though Slashdot has been
mentioned in the mailing list).

I.e. this has presentation impact and thus becomes un-microformatty because:
- it is prescriptive
- it doesn't follow examples in the wild

= Case 2 =

B has an existing hyperlink :
<a href="URL-B">this comment</a>

This happens in 40% of comments in the examples. It has been posited
that if URL-B is substantially similar to URL-A -- i.e. URL-B differs
by adding "#ID" to URL-A -- then we can do this in B

<a href="URL-B" rel="in-reply-to">this comment</a>

or as we'd see this "in the real world":

<a href="URL-A#ID" rel="in-reply-to">this comment</a>

and that we can say, by fiat, that when we see A.rel="in-reply-to" we
strip the #ID to get the real URL that the relationship is talking
about (i.e. URL-A, not URL-B).

Now, I have large objections to hacking URLs into pieces to understand
them better, but I'll let that slide unless this debate gets even more
details.

My primary objection is that this does not work because it is contrary
to the HTML definition of "rel" [1]: "This attribute describes the
relationship from the current document to the anchor specified by the
href attribute."

The current document, in the microformats context, is the comment in
question - that's a Rubicon that's long since been crossed. The anchor
specified by the href _is the comment itself_. So A.rel is asserting
an incorrect fact, that the comment is it's own parent!

Also note that once you start down this road, how do you deal with
nested comments, all of which have #ID tags!

And then we still have to add this markup to 60% of comments!

= Summary =

Using rel="in-reply-to" either:

- is prescriptive, requiring markup to be added to almost all comments
systems, OR
- requires URL hacking that:
-- still is prescriptive/presentational in 60% of cases
-- doesn't work in nested cases, if that's a design goal!
-- asserts incorrect facts, according to the definition of A.rel

Regards, etc...

[1] http://www.w3.org/TR/html4/struct/links.html#adef-rel

--
David Janes
Mercenary Programmer
http://code.davidjanes.com