"Dennis E. Hamilton" <> wrote on 12/17/2010
02:46:46 PM:
>
> I think there is another matter of importance to recognize here.
>
> The production of CSDs is cumulative, not modular. That is, every CSD
that
> is produced has to incorporate the (current form) of features in
previous
> drafts and be for all four parts of ODF x.y assuming the ODF 1.2
structure
> is maintained.
>
Certainly, to the extent we remain as a single, multi-part standard, all
parts need to go out for review together, even if any individual part has
not changed. This is not a huge problem. It just is more work in putting
together the review materials.
> Furthermore, the cumulative (or even incremental-only) change tracking
would
> become nightmarish since, presumably, each CSD in the track model would
> incorporate entire new features.
>
I don't see this as a particularly nightmarish. How is it different than
what we just did to put together the review materials for CSD 06?
Remember, each 15-day review is only incremental, showing changes back to
the previous public review. We don't need to show all changes back to CSD
01 (or whatever our initial 30-day review version was).
> I'm not sure how the subsequent public reviews in such a situation could
be
> reduced to 15 days, if we are grafting in feature modules at each cut.
>
The process says the review is for a "minimum of 15 days". So the TC
could vote to hold a longer review if there were additional review time
were required.
> Perhaps one might do effected parts only in some Public Reviews, but at
some
> point we are back to having to do a full review series to put a set of
> drafts on the Committee Specification and Standard track.
>
I think every public review will need to have the complete specification
under review.
> Maybe we need to take another look at modularization with supplements as
> well as the handling of extensions and the prospective movement of
> extensions from practice to standard.
>
I think modularization would improve things, but only if done right. It
is the old analysis question: how do you maximize intra-module cohesion
while at the same time reducing inter-module dependencies? Parts 2 and 3
are in relatively good shape. But Part 1 is still very monolithic. No
matter what we do process-wise, an 850-page specification is going to be
tough to edit, to review and to revise. I suspect that modularization
will also help implementors.
-Rob
> - Dennis
>
> -----Original Message-----
> From: [mailto:]
> Sent: Friday, December 17, 2010 10:27
> To: Patrick Durusau
> Cc: Michael Brauer; OpenDocument Mailing List
> Subject: Re: [office] Thoughts on ODF-Next
>
> [ ... ]
>
> On the other hand, a CSD/Public review does have overhead, both in the
> length of the review, but also on the administrative checks. Since the
> initial review is the longer review (30-days), there is an advantage to
> doing that as early as possible, to get that out of the way. Of course,
> once you've done that you have the burden of highlighting changes in
> future CSDs. But on balance I think we're better of to have a
CSD/public
> review early.
>
> [ ... ]
>
