I am having a problem with trackbacks on a weblog that has only weekly archives, not individual archives. What happens is that if another weblog links to a post, the discovered trackback ends up being the most recent post of the week, not the target post.

The root cause of the problem is the logic in Util.pm::discover_tb. This is the subroutine that scans the retrieved page for of the trackback target URL. Unless the find_all flag is set (which it is not for autodiscovery, see Entry.pm::discover_tb_from_entry::446) then the first matching RDF data is used. However, the RDF data is matched on both the full URL and the URL with the anchor stripped (line Util.pm::1002). Of course, every URL in the weekly archive has the same base URL (the URL for the weekly archive page), the only difference being the anchor.

For a concrete example, consider the post atThe archive page URL isThe permalink for every post in this week has that as the base URL, differing only in the anchor. In the RDF checks in discover_tb(), the variables url_no_host and perm_url_no_host will both contain the URL fragmentfor every RDF descriptor on the page. Therefore the logic will match the first post on the page every time.

Obviously, even if 6A fixed this problem (and I can provide code to do that, if 6A would be interested), I would still have a problem because I would need the other weblogs to upgrade and that's not going to happen anytime soon. I would like to change the way the RDF is set up as a workaround, but I don't see how to do that without some external support (such as MOD_REWRITE). Any hints or suggestions would be appreciated.

Thanks for the write up. I've submitted it to my team for review.