Anne van Kesteren

Encodings: US-ASCII and ISO-8859-*

Internet Explorer is the only browser that treats a resource containing all octets in the range %01-FF with the encoding label "us-ascii" in a special way. That is, it is the only browser that does not treat it as Windows-1252. Rather, when fetching such a resource over XMLHttpRequest all the octets in the range %80-FF appear to be dropped. And when fetching it directly in a browser tab they are converted to some unknown character, presumably U+FFFD.

Meanwhile with some of the ISO-8859-* encodings I noticed that Firefox (Minefield) has a difference compared to everyone else. The bytes in the range %7E-%9E are often mapped to U+FFFD rather than U+007E through U+009E. Not sure what that is about. Using them is invalid per HTML5, but so are other code points that are mapped just fine.