One of the issues with TrackBack currently is that the RDF that TrackBack inserts into pages breaks XHTML validation. Now, whether you should care about this or not is up to you--the fact is, the RDF is valid, and if the rest of your document is valid, then really, you've met the goal of validation, if not the present reality.
But the fact remains that we do want to fix this issue and we'd like to hear your thoughts on the matter. We are planning on adding xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns" to the <html> tag in the default templates for the next release; while the W3C validator still declares the document invalid, this seems to be the right thing to do. We do want W3C validation, though, so if you have ideas, let us know. Phil Ringnalda has posted a great explanation of the issue with links to further resources and (perhaps) a start to one of the possible solutions.
Here's the static output representing the TrackBack threading of this initial post as of 3:00 am PST.
This was generated by the web-based tool that we're still developing -- so, pay no attention to the ugly html that was generated (This of course, is the cue for people to view the source).
Note: There is a lot of circular TrackBacking going on because this is really just post to test the functionality and, a lot of people a pinging just for the sake of pinging (for this example, that's not a bad thing).
I visited a lot of sites today where people have been discussing TrackBack; while the general opinion has been very positive, there are a few users who have mentioned that, at this point, they just "don't get it."
This was totally expected. The reason? Interesting implementations of the feature require the critical mass that would be gained by other developers integrating the feature (or a variation that supports TrackBack) into their own tools.
A related reason is that we're all still at the novelty stage where the TrackBacking is done more for test purposes; again, interesting and successful implementations require context.
Interestingly, the method of using TrackBack that is being most discussed -- the ping between entries as a way to referencing individual posts -- evolved from TrackBack's original purpose: Category pinging.
Tonight (before I took a five hour nap), I was hoping to set up a section of movabletype.org to receive pings from posts related to certain fairly general categories. This too, would be more a test of the feature rather than a true contextual example. Why? Well, because it would be quite difficult to decide on what categories needed to be represented and how granular they should be. And, would we or should we expect users to set up their categories based on an imposed taxonomy?
Let's use posts related to Perl as an example:
Say (in a very abbreviated form) we set up category repositories in this way:
Topic: Perl
Category: Programming
-->Subcategory: Graphics
-->Subcategory: XML
-->Subcategory: Database
Category: Tutorials
Category: Core modules
Category: Security
Category: Perl 6 development
I'm assuming, for this example, that one would be using the category --> category pinging method; when using the entry --> category method on a case-by-case method, some of these categorization problems seem a little less evident.
So, what if in your set of categories, you just post about "Perl" -- that's your category; Where does it fit? If there was a general dump all Perl-related posts in a category such as "Perl: General," would that remove the value of categorization in the first place?
These are all questions that we will want to address as we continue to build TrackBack; though, I'm not quite sure there are any answers that we can all agree on.
In the vein of distributed taxonomies, I was thinking that a somewhat useful feature might be a way to import/export your category list. If you are the "master repository" of a certain topic and have compiled a fairly detailed category and subcategory set that readers (and TrackBack pingers) of your weblog might want to follow, then distributing the set would be built-into the system. And, maybe we can go even further with not having to manually export/import the list but somehow transfer between systems!
I'm getting ahead of myself.
For now, how about sending me lists of what categories you use and once I get enough and am able to pull out the most often-used categories, I'll compile some sort of test category pinging section on movabletype.org.
Depending on how granular your MT categories are and how often you may find yourself talking about TrackBack, you may want to associate this url:
http://www.movabletype.org/mt/mt-tb.cgi?tb_id=3
with a category related to discussion about TrackBack.
To do this, go to "Categories" and create a new category called "TrackBack" (or whatever makes sense to you). Then click on "edit category attributes." On this page, under outgoing pings, insert that URL into the appropriate field and save. Then, when creating a new entry that is about TrackBack, select the category that you've set up. That's it. From then on, you only have to set the category and save for the ping to be sent.
We're planning on adding a resources and tutorial section to movabletype.org; we'll also provide category ping urls for those who write often about MT hacks and tricks.
Note: This problem should only affect those who downloaded version 2.2 yesterday, before 10:00 pm (when we slipped in the fix into the distribution along with a fix that will prevent blank pings from reaching your site).
If, when using bookmarklets, the "TrackBack entries to ping" pull-down (or single line in the case of pages with just one TB-enabled post) doesn't show up, you may be seeing the same problem that The One True b!X noticed.
Ben writes about the technical reasons for the problem:
"The problem was that LWP::UserAgent, after fetching the page from the original server (www.movabletype.org/trackback/) then tried to parse the
section using the HTML::HeadParser module. But that module isn't included with LWP::UserAgent, so it was dying. So I read up on it, and I found out the way to turn off the automatic parsing. And so now it's automatically turned off when fetching documents."As I said earlier, if you downloaded after around 10:00 pm yesterday, this problem is already fixed. If, however, you downloaded it after this time, you'll want to download a new copy of the distribution and replace two files:
lib/MT/App/CMS.pm
This will fix the bookmarklet problem
lib/MT/App/Trackback.pm
This will prevent you from receiving blank pings
How it works
Be sure to read the documentation found in the manual to truly understand how TrackBack works. Here is a short and simple example of TrackBack in use (or course, minus the context of any real content)
This is just a regular weblog post with "allow pings" turned on. Imagine that I have written something terribly important, witty or enlightening.
Now, after reading this, you decide that you want to write your own post on your weblog that references this entry -- the normal sort of weblogging referencing that we're all used to seeing and doing.
Now, since you've decided to write or reference the original post and because you're eventually going to have to ping the site, you're going to need the ping URL of this post.
Currently, there are two ways to ping an entry. The first is to manually copy the ping from the TrackBack popup and paste it in the appropriate field of the new edit screen.
The second, much easier method involves using the bookmarklet. If you click on the Movable Type bookmarklet while on a page that contains a TrackBack-enabled entry, the bookmarklet will display a list of TrackBack-enabled entries. You can select the entry that you wish to ping from a pull-down menu.
Now, try to ping this entry.
Additionally, we have the category TrackBack set up in the right column of this blog. To ping this category, associate this url with a category called TrackBack:
http://www.movabletype.org/mt/mt-tb.cgi?tb_id=3
Note, you can call the category whatever you want -- TrackBack just makes sense for this instance.
Possible Uses Of TrackBack
We think that out of the box, the concept of TrackBack may be a bit confusing. It seems like the sort of feature that will start to make sense once MT users start to see real-life applications of feature.
But still, we've already thought of some examples where TrackBack can be used. It's certainly not just limited to pinging entries.
Multiple "authors" without author accounts
Say you want to have your readers contribute to your blog, but do not want to add them as an author; either because you want to limit the number of authors or you don't want the work of having to add new people each time someone wants to post something interesting. Or, you may not want their posts to "weigh" as much as your official set of multiple authors.
With TrackBack, you can set up a section of your site to receive pings.
Kristine, one of our beta-testers, used her site, The Red Kitchen, as an example:
"If I had a category named 'Red Kitchen Guests' and allowed pings to it... then anyone with an MT blog could post a recipe on their page and ping my guest category. Then it could automatically list a ping link and excerpt on the Guest category page."
Referencing Individual Entries
You write an interesting post. Someone wants to comment about your post or reference the link on their own blog. Instead of relying on referrals or a comment that says "I wrote about you," the person who commented would simply ping your entry.
As another beta-tester, Brenna, said, think about this an an extended comment that you store in your own database. It's a way to generate comment for *your own* site while continuing the conversation.
We've also developed a tool that will display a threaded view of the entries involved in a discussion across multiple blogs whose participants are using TrackBack. This will be a stand-alone tool, not contained in the Movable Type distribution. We should be releasing it soon.
Surveys & Question memes
On one site, a person provides a set of questions that he or she wants the weblogging community to answer on their own weblog (Think Friday Five, Monday Mission, Blogger Pride Survey). Once you have answered these questions, you usually return to the site and post a comment linking to your set of answers on your own blog.
Assuming that the question site uses MT & TrackBack, you can simply ping the original entry. On their site, your weblog name, permalink to the entry and possibly entry will show up in the TrackBack listing. All you had to do is save your entry.
Repositories of related content
Say you can't get enough of APIs. And, you want to start a repository that would be the greatest resource for people looking for discussions about the various APIs out there. So, you'd set up a repository page that would contain your categories and their ping URLs:
Google API
Amazon API
Blogger / MetaWeblog APIs
Those who spend a lot of time talking about these topics, would associate the ping URL to their own categories. So, every time they wrote about the Google API, they would simply select the category "Google API" (or whatever they call it) and press "save."
If someone knows about the category but only writes about it once in a blue moon, they could simple manually ping the category by using the entry pinging mechanism.