Re: ISSUE-4: Versioning Straw Poll

On Aug 31, 2009, at 19:28 , Chris Wilson wrote:
> I’d suggest a different wording, because I disagree that “…are not  
> useful in a web context and should be forsaken,” but I agree they’re  
> probably not necessary here.  I say this because the scope of  
> “limited deltas” that do not “break real-world code” are a highly  
> contentious subject, particularly given the intranet (and extranet)  
> application of web apps.

I chose the harshest meaning to make sure we were all on the same page  
— I'm happy so long as we have a resolution as to what to do within  
the boundaries of this WG (even if we disagree on the reasons why).

I can rephrase to a milder proposition though:

   This house believes that explicit version mechanisms on an API, such
   as have been done elsewhere using for instance hasFeature(), a
   version attribute on interface object, or a version parameter passed
   to a constructor are too brittle to be relied upon in a context in
   which multiple releases of multiple competing implementations must be
   supported in parallel since the failure of a single one condemns the
   entire approach.

   Future revisions should be strictly additive, or change names. In
   rare cases the WG may convince itself otherwise after careful
   consideration.

--
Robin Berjon
   robineko — setting new standards
   http://robineko.com/

Received on Tuesday, 1 September 2009 13:24:20 UTC