As Miranda noted in a comment to this post, Haloscan has an RSS/XML version of the most recent 10 comments available. If I save the XML file and then delete those comments, I’ll have the previous 10 comments and so on. Since I have 409 comments on Haloscan, I’ll have to repeat this process 41 times. Then I have to write a parser to merge all these files, excluding the repeated stuff. Unfortunately, Haloscan does not use the CAIF standard for its XML output. So another conversion script needs to be written. Then I can use Phil Ringnalda’s script CAIF2MT that he wrote for importing YACCS comments.
If I end up writing it all, I’ll put it up here for public use.