- From: Boris Zbarsky <bzbarsky@MIT.EDU>
- Date: Thu, 09 Dec 2004 22:15:01 -0600
- To: Chris Lilley <chris@w3.org>
- CC: www-svg@w3.org, ietf-types@iana.org
Chris Lilley wrote:
> IH> I would strongly recommend that specs disallow sniffing of any kind,
> IH> treating all metadata as authorative
>
> They do, although the xml encoding declaration is metadata too.
I'm sorry, but that's not what was being claimed about user-agents receiving
content with a Cotent-Type of image/xml+svg, and Content-Encoding or
Transfer-Encoding that happened to look like it was compressed (the argument was
about gzip, but I'm not sure it was ever claimed that gzip compression is
special in this regard, except insofar as support for it is required by the SVG
specification).
The claim was that under these circumstances, the UA should disregard the
server-provided encoding value ("identity", in the only way that it can be
provided), should decompress the data based on its guess as to what type of
compression was used, and should tell the user that it did so.
The futher claim was that this is consistent with all existing specifications
and best practices and was a desirable behavior.
These claims hinged on the claim that the magic numbers at the beginning of
common compressed formats are actually metadata. Per this argument, of course,
it's not clear why we have Content-Encoding and Transfer-Encoding headers at
all, but such was the argument.
Note that none of this has to do with XML encoding declarations or "charset"
parameter of the Content-Type header, and it's unfortunate that the two issues
ended up getting intermixed this way...
-Boris
Received on Friday, 10 December 2004 04:15:17 UTC