Recommendation on "The continued use of amendments to correct DRs"

[Rex Jaeschke]

Back in July of 2010, I wrote a paper, "Rethinking the Use of Amendments To
Correct ISO/IEC 29500" (WG4 N 0148). 

According to the Tokyo F2F meeting minutes, "Rex presented his paper, and
there was a lively discussion. Francis added insight into why we took the
path we did regarding having both a COR and an AMD. No decisions were made,
and need not be in the near future. Rex's main point was that he thought
that in future, DRs that propose or result in the addition of new features
rather than fixing bugs, should be closed without action, and re-submitted
as proposals for an AMD.

According to the Beijing meeting minutes, "Earlier this week, members
expressed interest in revisiting this topic. We discussed why we resolved
some DRs in CORs and others in Amds; we mentioned how some NBs reacted top
our criteria; and we looked at the kinds of issues we've closed since then
and that are currently under review. Action: Jesper, Chris, and Rex will
investigate this offline and bring back a recommendation on what, if
anything, we should change in this regard."

Recently, I circulated the following text to Chris and Jesper:

===============================

Let me summarize my thinking:

A COR is intended to be used to publish corrections to defects. An AMD is
intended to be used to publish new functionality. The problematic area is
when a correction to a defect adds functionality, as in the addition of a
new XML type, like we did in various DR resolutions.

That said, if the addition of functionality was done primarily to address a
DR (and not to bypass the usual AMD review process), I see no problem
putting it in a COR.

Publishing some corrections in a COR and some in an AMD means we have 2
processes going on in parallel with two different time lines and voting
processes. Each must have its own mutually exclusive schemas. And we might
not find conflicts between them until we attempt to merge them (as happened
this last time). And no official public combined schema ever exists until
all the documents are consolidated into a new edition.

The first time we did an AMD for corrections, quite a few members were
concerned about the public opinion on the "sweeping" and BRM-related changes
we'd made. However, we got NO comments/complaints at all. In fact, if anyone
outside WG4 opened AMD1 they would quickly say, "This is nothing but a large
set of corrections. It looks just like a COR and not at all like an AMD
(which has its own Scope, Conformance, Terms etc. clauses. There is nothing
to review here for those of us not directly involved in the project!"

The Directives require that we produce a new edition after we've published
two "updates", be they both CORs, AMDs, or one of each. If we retain the use
of AMDs for fixing DRs, then every cycle we'll reach that limit of 2 WITHOUT
having any real AMDs (like the ISO Date one and the other one from JP).

I believe that eliminating the use of AMDs to fix DRs will benefit WG4
members, the Project Editor, and those NBs that vote on our work. And I
don't see anyone being disadvantaged. WG4 has proven that it can do good
work, and that it can be trusted to manage its own business without having
to ask the joint balloting world to ratify our DR fixes.

===============================

Today, the three of us had a teleconference to discuss this issue. Both
Chris and Jesper are in agreement with my suggestion that we stop using AMDs
to implement DR fixes; that is, all DR fixes should go into CORs. And that
is our recommendation to WG4.

Murata-san, please add this topic to the Prague WG4 meeting agenda,
preferably early in the week, so members can discuss it in committee and
think about it overnight.

National Body members, please come to Prague prepared to discuss and,
hopefully, decide on this issue.

Rex