Re: [xhr] Appending a charset after an author added content-type

On 2013-02-19 19:51, Anne van Kesteren wrote:
> On Tue, Feb 19, 2013 at 6:44 PM, Boris Zbarsky <bzbarsky@mit.edu> wrote:
>> Note that Firefox adds the charset even if it wasn't set by the page because
>> web developers were asking for it to aid server-side processing...  The
>> alternative is that the server has no idea what to do with the data.  :(
>
> Does that include cases though where the page did set a Content-Type
> through setRequestHeader()? Because technically what Firefox does
> violates HTTP semantics as not every MIME type allows for a charset
> parameter.

Right. If other UAs don't do this, FF should avoid it as well.

Which reminds me about the funny requirement to rewrite charset 
parameters in place; this really needs test cases; if it's still only 
Firefox doing it it should be nuked from the spec.

Best regards, Julian