2015-05-21 4:11 GMT+02:00 Doug Ewell <[email protected]>: > Philippe Verdy wrote: > > URLs were initially deisgned to be stable (and this is still a strong >> recommendation). >> > [+ 559 words] > > It doesn't matter if they were designed to be stable. Users don't keep > them stable. > > I can't believe we're debating whether URLs are stable on a list where > people have raised concerns about whether 50 years is stable enough for ISO > 3166-1. >
I just say that the URL encoding itself is stable and allows to use them for stable references. The W3C itself uses URIs (in fact just URLs, even if they don't return a resource when queried) for making the XML schemas identifiables. In SGML there are similar stable identifiers (but in a naming scheme). In both cases they are meant to make identifiers unique and stable over time. An URL does NOT have to return a stable content, it JUST has to remain stable by itself. There's absolutely no obligation for its associated content to be accessible or retrievable. It will survive even if the referenced content is later changed or deleted: an URL is a valid URI, it is an identifier.
