Re: [apps-discuss] draft-saintandre-xdash-considered-harmful-01

[Peter Saint-Andre <[e-mail]>]

On 10/25/10 4:42 PM, Thomson, Martin wrote:
> On 2010-10-26 at 02:50:31, Eric Burger wrote:
>> What is the difference between X-mumble and Eric-mumble?  Does
>> private namespaces really help the fundamental problem?
> 
> Namespaces are the root of the problem, aren't they?
> 
> X- is just a (slum) namespace.  Nothing works here.
> 
> Eric-mumble is worse because Albert doesn't like Eric and so he
> duplicates the functions in his own namespace rather than cede
> control to Eric.  (BTW, that's an argument I don't see in the
> draft.)

Many arguments are not yet in the draft, and I shan't have time to
update it before the cutoff. But perhaps a discussion at the appsarea
meeting in Beijing would be productive.

> Better to make your extensibility simple.  How about having three
> tiers of "interoperability protection":
> 
> IP-C: Anyone can use a header with any name they choose.
> 
> Use at own risk.
> 
> IP-R: Anyone can register a header with any name.
> 
> Collisions are less likely, not impossible.
> 
> IP-S: RFCs* register headers with absolute exclusivity.
> 
> Safe to use.

Could you compare / contrast to provisional vs. permanent registrations?
The latter distinction might give us most or all of what we need
(depending on which problem we're trying to solve).

> Because the name isn't scope or qualified, headers can increase their
> interoperability protection status without breaking implementations
> that already use it.  This encourages the registration of headers
> with appropriate names.
> 
> The registry provides a limited guarantee of uniqueness.
> 
> * An standards track registration - or something similarly strong -
> causes existing registrations for the same name get swept away.  The
> review process can determine if this is appropriate.

Fun. :)

Peter

-- 
Peter Saint-Andre
https://stpeter.im/