So... I did a bit of a (tiny) rewrite to we_prepareScriptUrl(), which is 'the' function that did what I remembered: it takes the script URL and adds a question mark if not there, or a semicolon if a question mark is found.
So I renamed it to the more generic 'weUrl' and added a parameter to specify the original URL. If it's empty, it uses we_script. (Obviously!)
I'm a bit annoyed by the function itself... At some point it uses url.charAt(url.length - 1), when I could use url.substr(-1) (and save 6 bytes off the gzipped size...), but it's not supported in IE (even 10!). So I switched to url[url.length-1] but I remembered that we discussed this at some point and gave up on it because of poor support... However, I just tried in IE6 (emulated in a virtual box) and it frigging works...?!