Re: [apps-discuss] "X-" revisited

[Dave CROCKER <[e-mail]>]

On 7/11/2011 7:00 AM, Peter Saint-Andre wrote:
> On 7/10/11 7:18 PM, Mark Nottingham wrote:
>>
>> On 07/07/2011, at 6:53 AM, Dave CROCKER wrote:
>>
>>> (Entire topic:  X- was a good idea to avoid collisions with standards, but turns out to be a much worse idea for uses that become standards.  So, don't use X-".)
>>
>> Can we get this into the abstract... or make it the abstract?
>
> Wordsmithed in my working copy to:
>
>     Many application protocols use named parameters to identify data
>     (media types, header fields in Internet mail messages and HTTP
>     requests, etc.).  Historically, protocol designers and implementers
>     have often distinguished between "standard" and "non-standard"
>     parameters by prefixing the latter with the string "X-" or similar
>     constructions (e.g., "x."), where the "X" is commonly understood to
>     stand for "eXperimental" or "eXtension".  Although in theory the "X-"
>     convention was a good way to avoid collisions between standard
>     parameters and non-standard parameters, in practice the costs
>     associated with leakage of non-standard parameters into the standards
>     space outweigh the benefits.  Therefore this document deprecates the
>     "X-" convention for most application protocols.

I like the paragraph.

The specific re-casting of my terse "entire topic" statement is in the sentence:

>      Although in theory the "X-"
>>     convention was a good way to avoid collisions between standard
>>     parameters and non-standard parameters, in practice the costs
>>     associated with leakage of non-standard parameters into the standards
>>     space outweigh the benefits.

That looks fine, except that and I'd suggest saying evolution or movement, 
rather than leakage.

Whether the standardization is intentional or not, it's viewed as a Good Thing, 
whereas "leakage" is typically taken as a negative.

d/
-- 

   Dave Crocker
   Brandenburg InternetWorking
   bbiw.net