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

[David Janes]

On Mon, Nov 17, 2008 at 1:51 AM, Toby A Inkster <mail at tobyinkster.co.uk> wrote:
> David Janes wrote:
>
>> In both cases, consider a blog post A identified by URL-A, and a
>> comment B (on A) identified by URL-B.
>>
>> 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).
>
> Actually, that is not the case.
>
> I have gone through the list of examples and found that two of the example
> pages don't actually contain any comments, and two of the example URIs point
> to different portions of the same page (only one portion of which contains
> comments). Thus I have eliminated three pages from the list of examples and
> replaced them with slashdot, reddit and twitter to keep the list a nice
> round 25 pages long (easier for percentage calculations).
>
> Also, I've gone back over the examples Martin analysed, looking for
> interesting things he might have missed. For example, four of his examples
> include pagination links (comments are spread over multiple pages), but he
> didn't note that down. Two of the examples I added also include pagination
> links, bringing the total to 24%.
>
> On analysis, 16% of the examples *do* contain links suitable for hanging a
> rel="in-reply-to" link onto.

I wrote something that was factually correct, you then went and
changed the wiki to make what I said wrong. On all my previous posts I
had put in "as of this writing" for exactly this issue. Don't take
this as me objecting to adding new things to the list, just having it
asserted that I wrote something wrong when I didn't.

Comments on the addition Twitter below.

>
>> this has presentation impact and thus becomes un-microformatty because:
>> - it is prescriptive
>
> My proposal (that authors are free to use class="hfeed replies" and/or
> rel="in-reply-to") is non-prescriptive. Authors have a choice over whether
> to use rel="in-reply-to". If they don't want to, class="hfeed replies" will
> often be sufficient.

Just because they have a choice doesn't make it non-prescriptive. 16%
(as of 2008-11-17T3:00:05-400) is not a particular large number, and
as I'll point out below, it's lower than that.

>
>> 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>
>
> This is a horrible idea and I don't think anybody has suggested we go down
> this route. Clearly a blog comment cannot be in reply to itself. Now I
> understand what you meant when you were going on about URL hacking. But you
> seem to be arguing against a monstrous idea that nobody had proposed.

Actually, that is exactly what is being proposed here [1][2].

>
> Take a look at what I'm actually proposing, and it's nothing anywhere near
> as ugly as what you're arguing against there.
>
> My proposal is that all three of the following are allowed:
>
> http://buzzword.org.uk/2008/hatom-threading-1.html (class="hfeed replies")
> http://buzzword.org.uk/2008/hatom-threading-2.html (both)
> http://buzzword.org.uk/2008/hatom-threading-3.html (rel="in-reply-to")
>
> Authors may pick whichever is most appropriate for their page.

Either method has to stand independently on its own merits or fall
alone on their issues. Why wouldn't this be the case?

That 16% of comment systems (as of 2008-11-17T3:15:12-400) on the Wiki
could also be marked up using rel="in-reply-to" doesn't make a
compelling case to me *especially* since it's introducing seems
opening a much wider issue that should be fully discussed and analyzed
first [3].

In fact, it's not even 16%: Twitter is _not_ a commenting system and
certainly not a commenting system as outlined by the problem statement
[4]. Reply posts are not comments. Every post on Twitter is a free
standing post, some of which may be replies to other posts. This is
pretty well exactly the way people do threaded blogging discussions,
except twitter "makes up" the fact that you meant a post to be a reply
when you use @name [5] rather than allowing you to explicitly enter a
link.

If we want to make a microformat for replies, well, let's do that, but
let's make it a "reply" or "thread" microformat and make sure we cover
all the examples where that happens.

As mentioned in a previous post, I believe that the Entry Comments
proposal is sufficient and complete to cover all the examples given,
except noting Twitter above. That's not saying "wrap it as a
microformat", just that doing all the Wikiwork to complete the
proposal is is probably worthwhile.

Regards, etc..

[1] http://microformats.org/discuss/mail/microformats-new/2008-November/001935.html
[2] http://microformats.org/discuss/mail/microformats-new/2008-November/001936.html
[3] discussed in several e-mails were the general blog threading
problem and USENET at least once
[4] http://microformats.org/wiki/comment-problem - note the lack of
the words "reply", "technorati", "trackback", etc.
[5] as far as I can tell

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