Ah yes, the joys of URL encoding. Had to solve a similar problem in my own webserver software. Technically, it has to treat the URL in parts, encode them seperately and join the parts with the currect delimiters.

What is happening here is that the hash sign # gets encoded when it shouldn't be if i remember my HTTP specs correctly. This is a common problem and some browsers provide workarounds, some don't. Plus, some websites/webservers also have workarounds for some (though not all) encoding problems.

What's even more vaxxing about this specific issues is that part of the URL needs to be parsed by the webserver and a part of it (the #make_schema) is to be interpreted by the browser as a hint to scroll down to a specific line in the returned document.

I technically have the rights to fix this in the PerlMonks code, but i have absolutely no idea where the encoding of URLS is happening.