Re: XML Schema validation and https redirects

Michael Kay <mike@saxonica.com> writes:
> "Stable URIs" was one of the fundamental guiding principles of the
> web, and it was always going to be difficult to achieve when URIs
> incorporated the name of a specific comms protocol.

Yes. I mean the real shame, I think, is that the http: protocol wasn’t
originally designed with security in mind. (That’s not a knock on the
folks who designed it, the web literally didn’t exist yet and “before
the web” is by definition “before the web criminals”.)

And then, when security had to be added, it wasn’t practical (for
various technical reasons) to make it part of http:, it had to be a new
scheme. And here we are.

Looking back with our 20-20 hindsight, I think it would be easy to make
the case that namespace identifiers would have been better if they’d
been done like Java package names:

  <ex:someElement xmlns:ex="com.nwalsh.ns.example"/>

That would have had the same distributed naming properties as URIs
without any kind of protocol. It would also have enabled, if you squint
just right, <com.nwalsh.ns.example.someElement/> to be a EQName-like
form for the name.

But there were external constraints on the design space that couldn’t be
overcome. They turned out to be poorly conceived, more’s the pity.
But…like I said, here we are.

                                        Be seeing you,
                                          norm

P.S. For what it’s worth, the XML Resolver package has a flag that tells
it to conflate http: and https: when looking things up in the catalog.
So it’ll “just work” even if you get it wrong.

--
Norman Tovey-Walsh <ndw@nwalsh.com>
https://nwalsh.com/

> Curiosity never killed anything except maybe a few hours.