From a performance perspective, it seems it would make more sense to have the client load resources (JS, CSS, images) from / rather than a relative location. This way, the browser would be able to cache the resources properly and reuse them for calls and rooms. Also, the nginx configuration for loop-client standalone would be a lot more reliable if we go down this path. I understood there would be a tag of loop-client (0.10) tomorrow, and if this could be included in it that would make a lot of sense for us and avoid having hacky nginx configuration rules in prod.
needs investigation - enhancement. not clear how much impact perf or how straight forward - but if is impactful - take it.
I think we've gone beyond this now - the call urls are hardly used, so caching isn't going to help that much. wrt the nginx configuration, I'm hoping to remove call url support from the standalone in a cycle or two, at which point the configuration could be cleaned up. Although I've not checked, supporting full relative urls might actually make some of the ways that devs work (and the standalone node servers used by dev) more complicated, so I'm hesitant to do work on this, especially considering we could be removing the code soonish. Therefore I'm going to mark this as wontfix.