Re: The "deprecate/fixed-base" option

I think some default base for namespaces may be needed, to answer the
challenge of "relative to _what_, if the document is purely synthetic and
has no meaningful base URI". The main difference in htis proposal is that
we make that base "#FIXED" rather than just the default.

Downsides:

1)  I think this breaks the same folks who can't live with the Forbid
option. If MS is really using relative to point to schemas relative to the
document's base URI, having them point relative to a different base URI
wouldn't be a gain.

2) We still have to do all the computation for Absolutizing, and we still
have any open questions about whether URIs really are the right way to name
namespaces... but we get behavior close to Literal. Outside of the
political benefit of being able to say "see, we did absolutize", I'm not
sure this really gains us anything. If we're going to accept that overhead,
it seems to me, we might as well go all the way to Absolutize and let folks
learn the hard way that relative syntax is probably a bad choice.

______________________________________
Joe Kesselman  / IBM Research

Received on Thursday, 8 June 2000 10:29:42 UTC