Re: Banning relative - No real damage?

> What's more, we now have a real-live XSL example, dated 1999, using
> a relative URI.  As a dummy, yes, but it's a live document.

I posted that example to highlight that making an incompatible change
to an existing recommendation _will_ break documents and is not a
trivial step that can be undertaken lightly on the basis that
"No real damage" will occur. Taking the forbid option will cause
damage both to existing documents and to the reputation of the W3C as
a standards making body.

****however***** I did not post the example in order to argue that the
forbid option should not be taken. 

While forbid causes some damage, the damage caused by that option is
nothing compared with the damage that would be caused by taking the
absolute option, which would make relative URI namespace names legal
but unusable. While forbid affects some documents, the number is
finite (and probably smaller than first thought as I can't find any
Microsoft examples, despite earlier claims). The absolute
interpretation affects an unbounded number of future documents.

So if (as I suspect to be the case) the current literal interpretation
is unacceptable, and in particular is unacceptable to Tim Berners-Lee,
then forbid is likely to be the only option available.

Tim Berners-Lee probably objects to the suggestion that he has a
personal veto on this, but in practice I think he does. There are
going to be an increasing number of recommendations coming out
that build on namespaces, and if the core namespace recommendation
is something he finds architecturally unacceptable then he is in
an impossible position when it comes to signing off these documents as
W3C recommendations.

David

Received on Wednesday, 7 June 2000 05:04:54 UTC