When we parse README files, we have to convert certain characters in order to make them safe for display on the site. One example is converting > into >. Since your text includes that character at the end of your URL, the > is parsed as though it's part of the URL, which leaves the last ; in the text. Because of the security we must include for parsing READMEs, we can't change this behavior. To fix the issue in this case, either remove the > or make sure it is not directly next to the URL.
Michael Frauenholtz so you first htmlize the text and then look for urls to autolink in the html? eh...
If I have interpreted it correctly, the current behavior can break URLs that contain & too. Each & will be converted to & first and the URL will be terminated incorrectly after & leaving ; and following characters in plain.
After looking into this more, we were able to fix this. Originally we had separated our HTML escaping and URL parsing for performance reasons. Now, specifically in the case of READMEs that don't use any markup, we've determined that performance is acceptable enough to combine these for the desired output. Sorry for the quick decision to not fix.