[OpenID] Proposal: TrackForth

Lukas Rosenstock inbox at lukasrosenstock.net
Mon Feb 5 23:29:48 UTC 2007

Hi Dmitry

> Whenever you post a comment on somebody's blog you'd probably like to keep
> track of the responses. There are a couple of services that allow this
> (cocomments is one), but they use proprietary APIs and centralized
> management. Which is very bad in the context of decentralized identity.

I have already thought of this. To be honest, it was the reason that I 
included background transfer in my AddId!-draft, I was still thinking 
about it because it is somewhat different from all these other use cases.

> ...
> 1.  Alice posts a comment to a post on Bob's blog under the name "Alice"
> with the subject "Test" and the text "Hello, TrackForth". The permalink to
> Bob's post is http://bob.example.net/?p=12, and the URL of Alice's blog
> (provided in her comment) is http://alice.example.org/.
> 2. Bob's blogging system inserts the comment into the moderation queue and
> requests the page referred by Alice's link.
> 3. It scans the page for the trackback link tag, which it finds:
> <link rel="trackback" href="http://alice.example.org/wp-trackback.php" />

As you said later in the mail, this discovery mechanism could rely on 
Yadis, so that Bob only had to request an usually smaller XRDS document 
than a whole page (in TrackBack the blog needs to be checked if it 
really links a page in order to prevent spam, which is not necessary for 

> 4. It then performs the following request:
> http://alice.example.org/wp-trackback.php?title=Test&excerpt=Hello,+TrackFor
> th&url=http%3A%2F%2Fbob.example.org%2F%3Fp%3D12&blog=Alice

With that query it looks like a GET, but it's a POST, I think (should 
be, to keep HTTP design principles).
And is that wp-trackback.php in your example used for both TrackBacks 
and TrackForths? If yes, how does it determine the difference?

> 5. Alice's blogging system adds a new entry to Alice's comment list and
> returns a trackback success response.
> 6. Bob approves Alice's comment. The permalink to it is
> http://bob.example.org/?p=12#comment-123.
> 7. Bob's blogging system performs the following request: 
> http://alice.example.org/wp-trackback.php?url=http%3A%2F%2Fbob.example.org%2
> F%3Fp%3D12%23comment-123

Is it really necessary to make two requests? What about blogs without 
comment screening? Can they omit step 4 and just perform 7?

> 8. Alice's blogging system marks the comment as approved and requests the
> page containing Bob's post.
> 9. It scans the page for the Atom comments link tag, which it finds:
> <link title="Comments feed"
> rel="http://purl.org/syndication/thread/1.0/comments"
> type="application/atom+xml" href="http://bob.example.net/wp-atom.php?p=12"
> />
> 10. It then adds a new entry to Alice's OPML data.
> ...

This is out of scope of TrackForth, an extension based on standard 
protocols. Okay.
One might consider adding another "push mechanism" for notifying Alice 
about new comments in reply to her comment.


More information about the general mailing list