[Web-SIG] PEP 444 (aka Web3)

[And Clover]

On 09/16/2010 06:19 PM, Robert Brewer wrote:

> 2. Hardly anybody implements RFC 2047, and http-bis is phasing it out.

s/Hardly anybody/No-one/. Even if you wanted to, it's impossible to 
implement in any consistent way. The mention of RFC2047 is nothing more 
than an error.

RFC2047 is not on-topic as the top-level HTTP request/response entity is 
not defined in RFC822-family terms (HTTP uses its own grammar which is 
subtly incompatible). In header that might be able to fit an RFC2047 
encoded-word, no browser or server actually supports it, and the one 
place where RFC2616 actually references RFC2047 is in a quoted-string 
context, which RFC2047 explicitly states is not a valid place to use it!

This is why httpbis wants rid of it, and why Web3 shouldn't mention 
RFC2047 at all. There is no reliable mechanism today to get non-ASCII 
characters into an HTTP header, browsers treat non-ASCII header values 
differently and incompatibly, and all Web3 can hope to do is pass 
through the bytes unchanged without regard to what encoding they might 
represent.

> since folded and/or 2047-encoded lines are equivalent
> to their non-folded-nor-encoded variants, applications have no
> business emitting folded or encoded versions of these

Indeed. I'll go further: there is no place for header folding in HTTP, 
period - neither from the application nor the server/gateway. This is 
another feature httpbis deprecates. Folding is an RFC822-family trait 
that doesn't work on the web, due to poor server/UA compatibility and 
the existence of long, inherently non-foldable headers (eg. try passing 
a Authorization header containing a Kerberos ticket in 80 columns).

-- 
And Clover
mailto:and at doxdesk.com
http://www.doxdesk.com/