Automatic Resource Management, V.2

[Mark Mahieu]

2009/4/22 <brucechapman at paradise.net.nz>

> > At the end of the day I find all of
> > these
> > three solutions, try, other keyword, or scope, better than what we
> > currently have and therefore support any of the variations.
> >
> > -- Howard.
> >
>
> If language design decisions were based on accepting anything that was
> better
> rather than best, then you'd incrementally get way off course, because you
> cannot rework previous errors of judgement. As we all know, its hard enough
> to
> keep roughly on course when all practical effort is made to be "best" at
> each
> increment.
>
> It's called entropy and it never decreases, and the more you have, the
> harder it
> is to do more work. The wise thing to do is minimise the entropy of every
> change
> in order to maximise the opportunities for future change.
>
>
> Bruce
>
>
Indeed.

There are clearly many factors, but I think much of the debate around ARM
(and some of the other proposals) is interesting to look at from the
perspective of one particular hypothetical question:

"If the aim of Project Coin had been to identify just *one* proposal, but to
then allocate the same resources to it as are currently earmarked for the
entire project (in order to 'do that one thing right', so to speak), would
the proposal still be the same?"

Or, "just how big is this elephant anyway?" ;)

Mark