Re: Swamped (Was:Re: Call the question!)

I made this point in another thread, but I want to repeat it here since
it's directly relevant to the attempt to summarize the options and may help
folks understand why LITERAL won the straw poll.


The LITERAL option, as voted on by the Plenary's straw poll, is actually
LITERAL-and-deprecate.

That can be thought of as FORBID with a migration path. The intent is that
we agree that relative URI References are a bad choice as Namespace Names,
and should be avoided in the future, but that we need to allow people time
to phase out these names and switch to the alternative solutions we've
disussed.

In practice, the code winds up being identical with FORBID except that:

 (a) we don't actually prevent the names from using relative syntax (at
most we issue a warning rather than an error), and

(b) applications which wish to treat the namespace name directly as a URI
Reference AND want to support the deprecated syntax while doing so AND want
to see the absolutized form will have to generate it themselves.

There's still the question of whether/when the relative syntax graduates
from deprecated (it works but don't use it) to obsolete (support not
guaranteed so definitely don't use it).  The rough consensus was that this
change didn't have to occur any time soon; some folks felt it might never
actually become necessary. But the nice thing about LITERAL-and-deprecate
is that it allows us to proceed immediately and make that decision later.

______________________________________
Joe Kesselman  / IBM Research

Received on Wednesday, 24 May 2000 14:13:01 UTC