Re: [hybi] [Uri-review] ws: and wss: schemes

[Toby Inkster <[e-mail]>]

On 9 Sep 2009, at 01:43, Daniel R. Tobias wrote:

> Not really, to me at least.  That would be like saying that because
> you own that particular horse, you can declare its tail to be a leg
> even if the general community of experts in horses and biology in
> general have a definition of legs that it doesn't meet.

To call a horse's tail a leg would be wrong, because there is an  
established definition for what a horse's leg is. Taking a word which  
already has a meaning and recycling it to mean something else causes  
ambiguity.

Conversely, creating a new word to describe an existing concept that  
already has a perfectly good name risks confusion too.

But what we're talking about is a using a brand new name which has  
never existed before (URLs beginning with "http://websockets.net/" or  
some other currently unregistered domain name) to describe a brand  
new thing which also has never existed before (Web Sockets).

> Since I own the dan.info domain, [...] It doesn't, however, give me  
> the right or
> power to insist that the definition of the "http" scheme or its
> associated protocol doesn't apply to my URIs, and instead those
> particular URIs should be resolved using my own "FooBarBaz" protocol.

No, it doesn't.

It *does* however give you the right and power to *allow* your URIs  
to be accessed by non-HTTP protocols too, and to promote that fact.  
HTTP still applies to your URIs, but FooBarBaz does too.

HTTP 1.x is only one possible protocol that can be used to resolve  
URIs beginning with "http://". A user agent is free to choose other  
protocols instead. In practice, using HTTP 1.x is often a last- 
resort: most user agents implement a proprietary local disk based  
protocol to resolve URIs, only falling back to HTTP if that fails.  
(i.e. they implement caching.)

Software that understands your protocol will recognise the URI and  
handle it via your protocol. Software which doesn't will be none the  
wiser and simply attempt to handle the URI via HTTP.

-- 
Toby A Inkster
<mailto:[e-mail]>
<http://tobyinkster.co.uk>