Re: Motion 6
Maarten and P1788
On 24 Aug 2009, at 20:59, Maarten van Emden wrote:
My vote is NO.
Rationale: Motion 6 addresses an issue at level(s) below level
1 of the document accepted as per Motion 2, while there are
unresolved issues at level 1. One of these issues is the definition
of relational operations (at level 1, of course).
An important point is raised by Maarten. Must we resolve all level 1
issues before tackling any level 2 issue, and so on?
My view is no. I think we must (do our best to) resolve all RELEVANT
level 1 issues before tackling a specific level 2 issue.
My mental picture is: in some abstract Platonic world where our
standard already exists, there is a (hopefully acyclic) graph called
P1788, with level 1 issues being nodes at the top, level 2 issues
below them, etc. The edges are dependencies. Our real world group is
visiting the nodes in some order, and when it has visited them all --
lo, we have a standard!
The rule for traversal is that you may not visit a node till you have
visited all nodes it depends on, i.e., all its parents.
Maarten seems to be arguing for a strict Breadth-First Traversal. I
disagree. I think a bit of Depth-First Traversal is good:
- It keeps the momentum up and stops us getting bored.
- It helps ensure that the experts at all levels, from
abstract theory down to hardware, remain engaged.
- It rubs our noses in difficult issues of detail, like
the trade-offs between efficiency and generality, now
and in future, of which David Hough wrote on 6th August.
Or Maarten, am I misunderstanding? Do you argue that the issues in
motion 6 actually _depend_ on the level 1 "comparisons" issue? I
don't think they do. That's why I think they should be the subject of
a separate motion, and I hope I can persuade you to reverse your vote
on motion 6.
John