- From: Tim Bray <tbray@textuality.com>
- Date: Thu, 27 Mar 1997 12:35:14 -0800
- To: w3c-sgml-wg@w3.org
I feel obliged to address, at some level, the WG's concerns over the continued but temporary absence of a non-URL persistent naming facility in XML. Yeah, it's a problem. However, it must be noted that the Web, which visibly, often painfully, suffers from the same problem, seems to muddle along somehow. Someone has suggested that our URLs should be URIs. That's a good suggestion, and one we're looking at closely. I'm made uncomfortable in that URI's are kind of a speculative standard, just as, in 1986, were SGML (good) and the OSI network stack (bad). Many of us who voted, at the last gasp, against PUBLIC, really wanted to vote for it. But it came down to one issue: interoperability. Everything else in XML is either an example of something that has been proven to work (all the stuff from 8879, URLs, Unicode) or something that is an obvious solution to an easily-comprehensible problem. The naming problem is fearfully hard, and there is just no proven-to-work solution that we can steal. We *think* we have, in XML, defined a syntax such that any processor should be able to read any document. PUBLIC makes that go away. Sorry, but you can't sweep it under the rug, and I repeat: everything you can now put in an XML document has a *single* straightforward interpretation such that implementations should really interoperate first time, every time. With PUBLIC this is just not the case, and it was something that it was just too painful to give up. So, there are two ways out of the box: 1. Trot out a familiar, demonstrably implementable way to do lightweight solutions to the naming problem. We don't, in XML, ever require 100% solutions, but we do need something that it seems like large segments of the population will be ready to sign up for, or 2. Decide that the naming problem is just too hard to have any realistic short-term prospects for a single interoperable solution, but that naming is so valuable that we are willing to compromise our hard-won interoperability to buy it. Personally, I think that #2 is more likely to happen than #1. This story ain't over yet. Cheers, Tim Bray tbray@textuality.com http://www.textuality.com/ +1-604-708-9592
Received on Thursday, 27 March 1997 15:36:35 UTC