foolip

Inspecting the logs a bit closer leads me to believe that this is actually just spambots faking as IE which are too stupid to strip out the #respond out of the URL. These are all one-off requests to my server without any referer or previously loading any images or anything. So this shouldn’t be fixed, it should be left as is for stupid bots to trip on.

Adding the following to wp-settings.php fixed the problem, although I am not convinced that is is a good solution.

$_SERVER[‘REQUEST_URI’] = preg_replace(“/#respond$/”,”,$_SERVER[‘REQUEST_URI’]\);

Using a wildcard to remove all trailing #anchors is probably not a good idea as it would just hide potential future unrelated issues instead of showing up as 404 in the logs.

I see this problem in my logs as well and I believe it is a bug in some versions of IE 6 since it only happens for that UserAgent. I’m just guessing, but I think that if the # comes directly after a slash, IE doesn’t interpret that as an anchor but instead sends it as a part of the URI to the server. One solution would be ensure never to use this kind of URI, i.e. not putting / at the end of permalinks. The other solution might be to filter out the #anchor of the URL when it arrives to wordpress. This would give i.e. the correct page despite giving a stupid URL, although it would probably not jump to the proper anchor.