Posted by Alan J. Flavell on 09/04/05 15:23

On Sun, 4 Sep 2005, Jukka K. Korpela wrote:

> Indeed, but we know that John either wasn't referring to _any_ ASCII
> character or expressed himself very obscurely.

Right, but I was commenting specifically on the posting to which I
was following up. I think enough people had done their best already
to express the point that the original question was unclear, and no-one
is really sure what kind of answer it was looking for.

> It _is_ relevant that a non-ASCII character is involved, and less
> important what it is and how it has been encoded. After all, form data
> submission is defined for ASCII data only, when the default enctype is
> used.

That's an interesting point, indeed. As I found with my tests at
http://ppewww.ph.gla.ac.uk/~flavell/charset/form-i18n.html , browsers
don't feel themselves constrained to that particular rule, and typically
will submit %xx references to any 8-bit character from the character
coding in which the HTML page itself had been sent; but it had never
occurred to me to try the experiment from an HTML page whose coding has
been specifically set to us-ascii.

I've just looked out the test materials again, and repeated the test,
using us-ascii as the character encoding for the HTML page which contained
the submitted form; and I find that Mozilla submits the Latin-1 characters
in that illogical HTML-ified form which MSIE had introduced earlier, e.g
a single submitted character (ETH in this case) was submitted as
%26%23208%3B% (i.e the URLencoded representation of "Ð") [1]

Trying the same thing from IE6, you won't be surprised to hear that what
*it* submitted was %26ETH%3B , just for variety.[1 again]

I suppose I should write this up on the above page, when I get a spare
moment.

hope this is useful to someone, anyway


[1] I've no idea what kind of a cock-up Google Groups will make of
displaying this in its various formats, but I am posting plain text - if
GG chooses to interpret it as HTML, all bets are off. Use a proper
newsreader, folks!

[Back to original message]