Re: The problem of having multiple Content-Range headers in HTTP response

On Sat, Mar 6, 2010 at 2:50 AM, Yves Lafon <ylafon@w3.org> wrote:
> On Fri, 5 Mar 2010, Silvia Pfeiffer wrote:
>
>> Thanks for clarifying - it's good to read and understand these things
>> thoroughly. I was indeed wondering how Microsoft's scheme could work
>> when they in fact used commas.
>
> Btw "Cookie" is using a comma even if it is not permitted by the spec,
> requiring special handling and caring by proxies to be sure that there is no
> folding or split of that specific header.

You're talking about the Cookie/Set-Cookie header? Surely there are
many more other headers that have commas. Do only those that the
proxies have to deal with create a problem with commas?

Also, I wonder if Microsoft's Content-Range: rows
(http://msdn.microsoft.com/en-us/library/ee159574%28EXCHG.80%29.aspx)
has a similar problem.

Maybe then we shouldn't even use comma as a separator at all, even in
the URL? We might as well go straight to semicolon or some other
separator, then we can keep the syntax the same between URL and HTTP
headers and don't have to transform it, e.g.

http://example.com/video.ogv#track=subtitle;audio

=>  Content-Range: track subtitle;audio
and
      Content-Range-Equivalent: track subtitle;audio

What do people think?

Cheers,
Silvia.