Think this through for me, wudja?

I’ve had Seth Dillingham’s post about TrackBack, along with his ping to my entry and email, sitting around waiting for my brain to tell me it’s sharp enough to think things through for going on ten days now, and I’m starting to wonder if maybe instead of being sick I’ve just become permanently stupid. Could you please think it through, and tell us whether or not switching from (commented out) RDF to just a Technorati tag-style link for TrackBack autodiscovery would be a good thing? From what I’ve seen from the RDF-in-HTML Working Group, there will never be a direct solution that just lets us take the comments off, so it’s either GRDDL or whatever the (X)HTML Working Group is up to for XHTML 2.0 if we want to maintain some tenuous relationship with RDF, or just a link with a slightly hacky rel if we don’t.


Comment by Shelley #
2005-03-17 20:18:46

First, dump trackback!

There, now that this is out of my system, I am a little concerned about the use of the rel attribute for any and all ”want something in XHTML that validates” situations. If we continue along this path, we’re going to bring back BLINK, and I’ll have to jump off a cliff.

But the use of RDF for trackback autodiscovery never made much sense. After all, there is no ’meaning’ inherent with knowing how to have your machine talk back to my machine.

What is the value of knowing this information, this link to use for trackbacking? Is it of use in an of itself apart from its participation in a purely mechanical process?

I tried to play on this a bit when trackback came out, when I was still flirting with Threadneedle, but in the end the big piece that’s missing in the centralized pool of links, and Technorati provided that, not trackback.

Anyway, that’s my opinion for what it’s worth. Because I’m a woman in technology and by gol, I have an opinion!

(sorry, wrong topic. But I’m kinda on a roll.)

I hope you feel better soon. Will post you a photo of a purple crocus to cheer you up.

Comment by Jacques Distler #
2005-03-17 21:24:01

I never understood the commented-out RDF thang.

  1. Commenting out is fragile (as Seth notes)
  2. Almost all the information is redundant. The only new information that cannot be extracted directly from the XHTML is the trackback:ping URL.
  3. Why RDF?

A<link rel=”trackback:ping”> element which points to the … ahem! … related resource (the URL to ping) would seem much more logical. That’s what pingback autodiscovery uses.

I hope that doesn’t make Shelley jump off a cliff.

Comment by Phil Ringnalda #
2005-03-17 22:45:38

RDF because at the time it seemed like the forward-thinking web-standards-geek thing to use, when you wanted to associate some metadata with some resources. And while in the case of an unmuddled template for individual entry archives it’s true that you can extract all the information you actually need from XHTML, in fact parsing arbitrary (X)HTML for meaning is a nightmare, and in the case of anything other than individual entry archives, you have to tell the user for surely sure which URL to ping goes with which post. Thinking about the sort of output that I’ve seen people produce over the years as a result of unleashing their, er, creativity on perfectly valid templates, turning them into what Tag Soup becomes after it has been digested, I wouldn’t even consider trying to parse it without a full-on HTML parser that had been through the fires of being used by a browser, and even then, what semantics are going to remain to be extracted? There are twenty posts on the current page, and you’ve either got a URI with a fragment in the address bar, and you have to find something associated with that, or you don’t have a fragment, and you have to associate twenty titles with twenty TrackBack URIs. What would you rather parse, the mucked-with HTML, or some auto-generated RDF that the user can’t touch?

Comment by Anonymous #
2005-03-17 23:10:07

Jacques–no screwing around now, I’m a woman geek on the edge…

This is blink (will it work? Ah, bummer — it got stripped. No fun.)

How about adding /trackback on the permalink? Yeah, right. Now I’m just being a smartass. That won’t work.

Well, at least now we know, all forward thinking web standards people have found the truth of the semantic web, and who’d a thunk it could be found in the rel attribute?

Comment by Phil Ringnalda #
2005-03-17 23:28:11

Ah, sweet PATHINFO. Of course, over here in the dull world of static HTML, there’s no guarantee of anything up the path listening for it, but also, if I understood it right, IIS doesn’t give Perl the PATHINFO, and that was why we didn’t get to get rid of GET for TrackBack when we should have, because IIS had to use old-style URLs-to-ping, with mt-tb.cgi?tb_id=1 instead of mt-tb.cgi/1, which then makes even new implementations think they are dealing with an old only-knows-GET version, at which point I start thinking that <blink>ing chartreuse <h1>s sound pretty good, by comparison.

So very, very tired.

Comment by Phil Ringnalda #
2005-03-17 22:29:35

Dump it? Just because it’s a huge spam target, and most people have only ever gotten one or two TrackBacks in their life (and those probably were just automatic ”I linked to you” pings, not really TrackBacks at all)?

Yeah, probably right, though it won’t just die clean, it’ll drag out, with people using it one-way, and others wondering over and over again why you don’t use it, and even fewer using whatever you replace it with than ever used TrackBack (well).

But I do love me some crocus. The only thing more eager for spring than me.

Comment by Geof F. Morris #
2005-03-17 21:28:56

Well, but couldn’t you dynamically generate a meta element for this? Maybe I’m too stupid to know why the autodiscovery has to be inside the body element rather than the head, but I see syndication autodiscovery done in the head, so my flighty aero-eng mind says, ”Oooh! Do it again!”

[Personally, TB autodiscovery freaks me out a little, mainly because I’m a knuckle-dragging WordPresser who’s used to having the URI forcefed to me.]

Comment by Phil Ringnalda #
2005-03-17 22:24:54

Well, it goes a little like this: when TrackBack was first introduced, back before they added in autodiscovery in the sense of ”just spew pings at every URL you link to in any post, because otherwise people won’t shut up about Pingback,” there was only autodiscovery in the sense of ”go to someone’s post you’re going to respond to, click your posting bookmarklet, and we’ll pick up the title, your selected text to quote, and also the TrackBack URL.” That was some good stuff, and had it been restricted to the only places where it makes sense, archive pages, we would have been fine: rather than embed RDF, it could link to a separate RDF file, that would have amounted to lovely useful rich and tasty RDF metadata that could actually be parsed by RDF parsers, without them having to know about commented out flup, and having to use a regex to extract stuff from comments before parsing it. But the beta testers wanted to see a select menu of choices of things to ping show up when the hit their bookmarklets on a blog’s main page, and rather than explain to them that they were being foolish, because you don’t respond to a particular post by starting out on the main page, and picking up a link to the main page, you start at the post and get the link to the post, Ben caved. That was a problem, because Movable Type doesn’t have any idea what posts are included in an index template. Individual archives, sure, monthly archives, it can figure it out, anything that can be a permalink target also has a defined set of posts, but a index template might include any sort of selection of posts: the last seven posts, the last five days, the last two plus the five most recently commented-on posts, it just doesn’t know, until it builds the page. So that’s why we’re stuck in the <body>: MT can’t do it in the <head> of a page where it serves no useful purpose.

Comment by Roger Benningfield #
2005-03-18 06:10:07

Phil: Why couldn’t they just loop over the entries twice… once in the head, and again in the body?

Comment by Aristotle Pagaltzis #
2005-03-21 16:44:43

But how do you know which part of the template to build during the first loop and which during the second? If you do the obvious thing and make it two templates, how do you explain to users why that is and how they should set up their templates?

(IMHO the problem is with all those template-driven tools that just glue together strings and will happily produce tagsoup, completely oblivious that someone downstream expects a document tree. But I digress.)

Comment by Phil Ringnalda #
2005-03-21 18:59:16

I think if I were crazy enough to try doing it, I’d run through the template once with a wrapper around, um, something, either all the entry-producing containers or the entry method itself, looking to only record which entries were touched on the first pass, and then actually build the HTML on the second pass. But as that handwaving shows, I don’t have what it takes to do it.

I’d love to play with some sort of DOM-based blogging tool, but what are you going to write it in, that will give you a decent sized potential market? Python? My host provides a dusty 2.1.3, which is more than most hosts do. PHP 5.x? When will it have fairly wide distribution? Some Perl module? How many hosts have it installed?

Comment by Arve #
2005-03-18 02:59:44

Having seen both Opera, Firefox and Internet Explorer having rendering bugs triggered by where in the document I place the commented-out RDF Data, I vote for getting rid of it.

Comment by Seth #
2005-03-19 04:15:13

Arve, I don’t think it triggers any rendering bugs. The bug is in our understanding of how comments are supposed to work, as I explained in the message Phil pointed to.

Comment by Arve #
2005-03-19 05:25:30

I’m not simply refering to the fact that SGML comments end with ’’, I’ve seen rendering mess up with complex/evil CSS layouts be triggered by the presence of perfectly valid examples of trackback RDF data in SGML comments as well.

While parser errors certainly are the responsibility of browser vendors, it still is an argument for going to autodiscovery link constructs instead.

Trackback by truerwords #
2005-03-18 07:18:51

Phil Responds (sorta) re: HTML Comments and Linking Technologies

I thought Phil was ignoring me. I wrote to him on the 7th, after posting about HTML comments and embedded RDF , to ask what he thought of my suggesting regarding invisible links pointing to autodiscovery documents. He wasn’t ignoring me, he just couldn’t

Name (required)
E-mail (required - never shown publicly)
Your Comment (smaller size | larger size)
You may use <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <del datetime="" cite=""> <dd> <dl> <dt> <em> <i> <ins datetime="" cite=""> <kbd> <li> <ol> <p> <pre> <q cite=""> <samp> <strong> <sub> <sup> <ul> in your comment.