Ok, this is sweet. Short story for the impatient: CoComment is now fully supported on this blog, and everything works the way it should! :)

Long story for the curious: The issues that I described in my previous post were due to one basic fact - there was too much guesswork involved in harvesting the necessary information. Different blogs and blogging engines have different naming schemes and ways of passing data around. Coming up with a way to selectively extract information based on the type of blog is tricky, and even then, it may pick up incomplete or incorrect information if the author has made some changes to their template, as we noticed.

Eliminating the guesswork, as you might guess, makes the process much more reliable. Merlin from the CoComment team decided that one way to do this would be to ask blog authors to explicitly define a set of variables in a JavaScript block that contain most of the information the CoComment bookmarklet needs to complete its task successfully. Of course, the values for these variables can be dynamically generated using PHP, ASP etc. For more information on how to implement this, take a look at this post on the CoComment TeamBlog.

So basically, what did I do to make this blog work with CoComment? First, I had to change the naming convention of the comment form to match the one used by TypePad, which is just one of the many blogging engines already supported by CoComment. That was what I did yesterday to get the basic functionality working. Now to fix the "untitled" problem and the clumping issue, I had to explicitly define variables that store the blog title, the post title, and the complete URL, including the unique post identifier, so that the bookmarklet could extract this information. Since the guesswork had been eliminated, everything worked perfectly. :)

So go ahead - leave a comment, and test it out if you want.