[Standards-JIG] JEP-0001: Draft Versus Final

Peter Saint-Andre stpeter at jabber.org
Tue May 24 18:07:15 UTC 2005


On Mon, May 23, 2005 at 05:28:39PM -0700, Chris Mullins wrote:
> PGM Wrote:
> 
> > Making them mandatory after a JEP is already Draft 
> > should not be allowed. 
> 
> Section 8.1 of JEP 1 doesn't say anything about that. It only talks
> about "rough consensus" for a JEP to be moved into Draft state. 

JEP-0001 does not specify exactly how to handle feedback received from
implementation experience gained while a JEP is in the Draft state. It
certainly does not speak to the particulars of JEP-0060. In general, we
seek rough consensus and running code. Discussion does not end when a 
JEP has reached Draft, and Final is not simply a rubber-stamp of the 
Draft JEP. As to how we currently handle changes to Draft JEPs, the
following seem like some rules of thumb (which we could change or 
codify in JEP-0001):

- Get it right
- First, do no harm
- Fix what needs to be fixed 
- Don't make wholesale changes
- Don't add major new requirements
- Address feedback from implementors
- Gain Council approval for all changes
- Follow the protocol design guidelines (JEP-0134)
- Don't break existing implementations unless necessary

> Clearly, for PubSub, we have Rough Consensus, but defiantly not
> Unanimity. 

I don't see anyone getting defiant about the lack of unanimity. Granted,
some people like their consensus rougher than others, but I don't see
any need for defiance. ;-)

> > We'd be breaking exsting implementations :( 
> 
> That's what "Final" is for, is it not? After reading JEP-0001 it seems
> that "Draft" is a good time to make changes. 

It's a better time than when the JEP is Final.

I don't know that we have much of a meta-issue here. Yes, it would be
good to codify and define some best practices for moving from Draft to
Final, and we can do that in parallel to the JEP-0060 discussions if
desired. But if we use some common sense, I think we can work out the
JEP-0060 issues separately from the JEP-0001 issues.

/psa




More information about the Standards mailing list