Re: TIBCO objects to last call

Jonathan Marsh wrote:
> Anish wrote:
> 
> 
>>Oracle is sympathetic to Tibco's concerns, specifically we agree that:
>>1) Some critical issues have been closed hastily in a hurry to go to
>>Last Call.
> 
> 
> David's proposal would result in an on-the-wire change of adding one
> attribute for the purpose of identifying extensions.  (An earlier
> proposal also proposed changing a couple of element QNames and coining a
> new URI).  These changes are modest.  We discussed them for several
> weeks which seems commensurate with the scope of the changes.  I object
> to your characterization of this process as hasty just because you
> didn't like the outcome.
> 

To be clear, not all proposals required a on-the-wire change.

If you look at the straw poll [.1] option #4 got 9 votes. Option #4 + 
option #1 (which are in the same bucket) got 11 votes whereas option #2 
+ option #3 (which are in the same bucket) got 14 votes. Option #4 was 
not fully developed. We could have proceeded by provided more time to 
fully work out option #4 and see if that made a difference. In a typical 
"non-fast-track" W3C WG that is how it would have happened, essentially
getting us closer to consensus.

[.1] http://www.w3.org/2002/ws/addr/5/03/21-ws-addr-minutes.html#item07

> 
>>In our opinion the unreasonable schedule specified in the
>>charter (the so called "fast-track" approach) is to blame for this.
> 
> 
> You are entitled to your opinion.  I'm entitled to mine.
> 
> It is my opinion that your opinion above is misguided.  Tibco joined the
> group late, and raised these issues after the time the WG was scheduled
> by the charter to go to LC.  This could have happened whether a schedule
> is compressed or not.
> 

The comment was not limited to issue i054. The opinion was based on 
observations made throughout the life of the WG. The schedule 
constraints seem to weigh so heavily on the WG that proposed resolutions 
(whether they are better designs decisions or not) that may require the 
schedule to be disturbed are often met with stiff resistance on the 
basis that there is no time to do the job properly and not on technical 
grounds. Not the best way to create a foundational Web services 
specification, IMHO.

> 
>>2) Going to Last Call should not be a goal in itself.
> 
> 
> Neither is delay for the sake of delay, which is what I believe you are
> promoting.
> 

This is not the case. Oracle would like the WG to proceed to 
Recommendation, not just Last Call, as expeditiously as possible. But if 
going to LC now is *the* objective then it would be hard for us to side 
with the subsequent argument -- lets not make a particular change as it 
will require us to have another LC period.

Oracle has always believed that WS-Addressing is a foundational spec, 
and therefore it is critical that we get it right and in a timely 
fashion. An apt quote that was heard in this regard at the recent W3C 
plenary in Boston:
"WS-Addressing is to Web services as URIs are to the Web."

> 
>>If concerns are
>>expressed whose resolutions may result in changes that require us to
>>go
>>through yet another Last Call, then we should examine our decision to
>>go
>>to LC carefully.
> 
> 
> I reject the implication that Microsoft's position was not the result of
> careful consideration of the issue over a period of weeks.  I have no
> evidence to suggest anyone else's position was uninformed.  You should
> back that implication up with proof.
> 

Case in point: option #4 was not a fully developed proposal and it 
received 9 votes in the strawpoll.

> 
>>We would rather see one LC period for the specs rather
>>than multiple such LC periods. The concerns laid out in David's email
>>regarding the ramifications of MAP extensibility (or lack thereof) are
>>significant enough that it may change the specifications in ways that
>>may require another LC period.
>>These concerns are worth considering
>>and
>>are a result of the particular resolution adopted by the WG for issue
>>i054.
>>We do appreciate the fact that the WS-Addressing specification(s) is
>>(are) needed urgently, but we believe that WS-Addressing is very
>>fundamental to Web services infrastructure and it is worthwhile to
>>spend
>>the time to get it right.
> 
> 
> Your posturing assumes that the status quo is sub-optimal.  We (and
> apparently others) do not believe that and voted accordingly.  I don't
> see any new information to warrant weeks of further discussion before
> holding a formal vote to move to Last Call.
> 

I would like to suggest a compromise to move us forward to LC. This may 
perhaps make folks who are concerned with issues (raised but not 
officially logged yet) arising out of resolution of issue i054 happier, 
and at the same time allow us to go to LC now.

The suggestion is to add some explicit text to the LC draft highlighting 
the issue and soliciting feedback on a priorty basis from external as 
well as internal reviewers and proceed to Last Call.

I will send out a subsequent email with details for this proposal.

-Anish
--

Received on Saturday, 26 March 2005 02:35:01 UTC