W3C home > Mailing lists > Public > ietf-http-wg@w3.org > July to September 2009

Re: draft-reschke-rfc2231-in-http-02

From: Daniel Stenberg <daniel@haxx.se>
Date: Tue, 4 Aug 2009 14:10:42 +0200 (CEST)
To: Julian Reschke <julian.reschke@gmx.de>
cc: HTTP Working Group <ietf-http-wg@w3.org>
Message-ID: <alpine.DEB.2.00.0908041352400.3759@yvahk2.pbagnpgbe.fr>
On Tue, 4 Aug 2009, Julian Reschke wrote:

> - RFC1867 is obsoleted by RFC 2854

Oh, thanks, I hadn't noticed. But I think would rather say that technically it 
has been obsoleted in parts by RFC 2854 and in parts by RFC 2388.

> - RFC 2388 seems to be a bit confused about what encoding to use, it 
> mentions both RFC 2047 (which I think is a bug), RFC 2184 and RFC 2231 (2231 
> obsoleted 2184) -- I think this should be addressed with an erratum (Larry, 
> are you reading this...?)

Right, but no matter which MIME specs it should refer to it considers the 
headers to be MIME (as opposed to HTTP-like). Thus it seems RFC2231 _should_ 
be our guide.

> - the Apache commons fileupload library doesn't seem to attempt to handle 
> either RFC 2047 nor RFC 2231 encoding

So how does byte codes within file names above 128 get included? Or are those 
bytes just skipped?

> - IE appears to send non-ASCII characters as raw ISO, at least if the source 
> page used that encoding,

This is similar to what curl currently does and what initiated my 
investigation into this area. It simply has no notion of encoding and passes 
on whatever the user pass in as "file name".

Someone detected that curl thus didn't do the upload the same way one of the 
browser did. Now I realize it will be hard to mimic those without getting my 
brain to hurt...

> - Firefox appears to escape with &#ddd; notation.

I have a hard time to see how HTML encoding can be the right thing here.

-- 

  / daniel.haxx.se
Received on Tuesday, 4 August 2009 12:11:35 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Friday, 27 April 2012 06:51:08 GMT