Should protocol setter remove tabs and newline?

[TimothyGu]

Consider the following test:

u = new URL("https://github.com/whatwg/url/issues/new");
u.protocol = "ht\ntp";
u.href

On Chrome, Firefox, and Safari, u.href retains the original https protocol. This behavior as it relates to the Location object is in fact tested by WPT(!): search for \x0A in html/browsers/history/the-location-interface/location-protocol-setter.html.

[domenic]

I don't have a strong feeling here, but I lean toward keeping protocol consistent with all other invocations of the URL parser and getting all browsers to update. It seems pretty gross to have a special carve-out for one particular state override.

On the other hand, when all three browsers do something, probably the spec should be updated to match it, even if that thing is really weird.

I'm curious what @achristensen07 thinks...

[achristensen07]

Given that all browsers agree on behavior, I think we should change the spec to match behavior in this case

[achristensen07]

I guess I don't have a strong opinion here. I don't think there will be a lot of compatibility issues if we change this.

[achristensen07]

Should we at least update the tests to be consistent with the current specs? I also found that http://wpt.live/html/browsers/history/the-location-interface/location-protocol-setter-non-broken-weird.html and http://wpt.live/url/url-setters-a-area.window.html also contradict each other.

[domenic]

Yeah, we should definitely update the tests to be self-consistent, and probably it's best to align them with the current spec. (Unless we want to update the spec, which I think everyone is OK with, but nobody is insisting on...)