Let me first say that my opinion on using ember has actually changed in the past month, since I've filed this issue on your repo. Back then I was in the "No ember at all costs" camp, because I was convinced that using js to generate pages is a bad idea in any case. Since then I have started looking at the issue from multiple angles and shifted my opinion to be more differentiated, to be concerned about the original reasons why I held that opinion, which is that I didn't want regressions in performance.
That was why I put "Don't regress on performance" onto my wishlist instead of "Don't use ember".
Blanket arguments of the form "Ember is JS, JS is bad, so using Ember is bad" are not constructive, and are especially bad when they come from people who won't contribute but only make demands.
That doesn't mean that all concerns of those people should be ignored. I think performance concerns from potential users are legitimate, and you and @carols10cents seem to agree that performance is important. This is great!
I stand with my opinion. I'm fine with ember or whatever web technology you or anyone else working on rustdoc are most comfortable with, as its not my position to dictate you a technology. But I wish there would be no regression on performance, because that's something I will be exposed to in the future.
I was the person who was originally doing the comparison with crates.io in this thread, so I guess that's addressing my comments as well.
I was bringing up crates.io because I'm not very familiar with ember, and it was the first site where I knew I was using ember when visiting it. So naturally, I took crates.io for comparison. I think its similar for other people on the thread. The concern I put on my wishlist was "crates.io is slow, please don't make rustdoc as slow as crates.io, please keep it at least as fast as it is right now". Its not "crates.io is slow, so ember is bad, don't use ember for rustdoc".
As a closing note, I'm pretty happy that various people have announced to create alternative rustdoc tools, because in the end there is no better way of finding out which way of solving it is best than by trying it out.