It currently performs a case sensitive one: <http://searchfox.org/mozilla-central/rev/59cd73fbfa14384a81a4e3eb17d3881b372702c4/netwerk/base/RustURL.cpp#318>.

Pushed by eakhgari@mozilla.com: https://hg.mozilla.org/integration/mozilla-inbound/rev/239f0df955d1 Perform a case insensitive string comparison in RustURL::SchemeIs(); r=mcmanus

The scheme of an `Url` object is always lower-cased by the parser, so if `SchemeIs` is only used with literal strings that are also lowercase, then the case-sensitivity of the comparison doesn’t change anything.

I'm with Simon, are we sure this was a good change? Or do we have legacy callers that don't provide lowercase input?

Ehsan, was there a specific bug that motivated this change?

(In reply to Simon Sapin (:SimonSapin) from comment #3) > The scheme of an `Url` object is always lower-cased by the parser, so if > `SchemeIs` is only used with literal strings that are also lowercase, then > the case-sensitivity of the comparison doesn’t change anything. There is nothing mandating what kind of strings can be passed to SchemeIs. But yes, if you pass a lower case string to that function then this patch won't change anything. The test that landed with it should have made this quite clear. (In reply to Anne (:annevk) from comment #5) > I'm with Simon, are we sure this was a good change? Yes. Any implementation of nsIURI needs to adhere to the same API contract. If we think that an API contract needs to be changed, such a change needs to be made to all nsIURI implementations. RustURL doesn't get to implement its own semantics. (In reply to Simon Sapin (:SimonSapin) from comment #6) > Ehsan, was there a specific bug that motivated this change? No, I found this during an audit of the code looking for an unrelated issue.