Cart
[Home] [By Thread] [By Date] [Recent Entries]
Understood. Many folks come to that conclusion at some point. So, it comes down to "use a URN" except URNs allow the protocol morph and that starts the problem all over again because names, locations and identifiers are conflated in the WWW Holy Writ. That is why I pointed out the Public ID. It's legal and it may be of some benefit to have owner, description, version information in there (authoritative contract). The argument comes back, "but I MIGHT want to resolve it and I hate catalogs" and the mess starts all over again. So, this comes down to the cost of preserving that option. If you want to avoid it, choose and then accept that by that choice, things like RDDL have to be gotten by extending the catalog. We've always known it's a wart. Would you rather process a SUBDOC? len From: Seairth Jacobs [mailto:seairth@s...] Hmm... I missed this one. Well, that would certainly be a good way to do it. However, I was just suggesting that if someone wanted to use a URL to identify a namespace, just use a scheme other than the "http". To avoid ending up with a proliferation of new scheme names, we could just define a common one that everyone can use. Whether the URL would be helpful in locating schema documentation or not is another question. The main point is to avoid the use of the "http" scheme when using the URL format, which would at least reduce the confusion caused in situations like "http://www.govtalk.gov.uk/CR/core".
|
