Re: [proto-team] Re: PROTO - proceeding on adding PROTO shepherds to the tracker

Henrik Levkowetz wrote:
> 
> on 2006-05-15 20:51 Lisa Dusseault said the following:
> 
>>On May 14, 2006, at 7:49 AM, Henrik Levkowetz wrote:
>>
>>>3.2. Notifications	
>>>
>>>   Under the assumption that the tracker already has the ability to  
>>>send	
>>>   email notifications when certain state transitions occur, the  
>>>tracker	
>>>   should be able to send out notification to one or more email	
>>>   addresses when one of the followings states or sub-states are	
>>>   entered(R-012):	
>>>
>>>   * Awaiting MIB Doctor Review	
>>>
>>>   * Awaiting Cross-Area Review
>>>
>>
>>It seems that there are other kinds of expert review besides MIB  
>>doctor.  You could consider the security review to be an expert  
>>review although naturally that can also be cross-area review.  The  
>>MIME Types list does some kind of expert review which could be  
>>handled the same way as MIB doctors. And it's interesting to consider  
>>expert review for i18n issues.  Should this be simply "Awaiting  
>>outside review" or, if we want to preserve the difference, "Awaiting  
>>expert review" and "Awaiting cross-area review"?  If we preserve the  
>>difference, what if both kinds of review happen simultaneously?
> 
> 
> Yees, I've already been pondering how to capture the possibility of
> more than one review being pending at the same time.  Either we can
> reduce the state to something very general, or we can consider ways
> to either have multiple instances of a sub-state, or having a sub-
> state with an annotation filed.
> 
> Unfortunately, both of these are bound to result in larger extensions
> to the tracker's capabilities than what was envisioned for this
> round...
> 
> I think that with no constraint on the amount of change we could
> introduce I'd advocate a fairly complete set of explicit review
> actions, and the possibility of setting more than one sub-state
> at the same time.  Then it's easy for other tools to pick up the
> sub-state and use that to trigger other actions.  Any free-form
> field makes that much harder and less predictable; and reducing
> all the states to very generic ones makes it impossible to kick
> off a particular review dispatch from the state change.
> 
> Whether we have the option to include this in the work package I
> don't know, however.

I would hate the result to be that the tracker artificially
serializes reviews.

I think that you should include the requirement, but with
flexibility for the implementor. If handling multiple parallel
review-waits is a challenge, the fallback as you say is annotation.
(You already have "5 reviews" as a substate. I was wondering
how that was going to be handled, except by annotation.)

     Brian

> 
> 
>>I do have an action item from the retreat to work more on the early- 
>>review problem, so my questions here are somewhat influenced by  
>>thinking how that might work.    (And by the way Allison, Aaron,  
>>Henrik and proto-team, I welcome input from all quarters on that.)
> 
> 
> Noted.  You should know (you probably do, but I mention it anyway)
> that Tero Kivinen has started working on a support tool for review
> groups.
> 
> 
> Regards,
> 
> 	Henrik
> 

_______________________________________________
proto-team mailing list
proto-team@ietf.org
https://www1.ietf.org/mailman/listinfo/proto-team