[
Lists Home |
Date Index |
Thread Index
]
Tim Bray wrote,
> Miles Sabin wrote:
> > At the risk of stating the obvious, the only constant on the web is
> > change. An "axiom" which states the opposite is just plain daft.
>
> Agreed. We're talking best practices here. And we are learning how
> to design URI spaces for longevity (hint: put dates in the URIs).
> There are grounds for optimism... -Tim
This helps enormously, I agree.
But solutions to the versioning problem don't come quite that easily.
Sure, when you move from your ../2000-09-01/.. URI space to your
../2003-03-17/.. URI space you can _try_ to leaving forwarding pointers
from the old space to the new space. Sometimes that'll work, but
there's no guarantee that it will in general. In any case as time
passes you'll be left with an ever increasing maintenance burden of
retired URI spaces. At some point it has to be reasonable to let the
ancestral ones die.
So, at a cost and imperfectly, this puts the problem off. If you're very
lucky and you can wait for stale references to disappear the problem
goes away altogether. But this is all a far cry from "cool URIs don't
change". TBH, to me it looks more like the antithesis of that: in fact
it's the cool URIs which _do_ change, and it's the stale old ones which
stay the same.
Cheers,
Miles
|