On Wed, Feb 09, 2005 at 12:15:38AM +0100, Florian Weimer wrote:
* Rick van Rein:
Back then, the URI *identified* a resource, but this doesn't make much
sense in the OpenPGP context (were fingerprints are used to identify
keys). The keyserver location is just that, a location.
The URI terminology has changed since then.
If URI doesn't mean what it used to, does it then make sense to change
both the keyserver and policy URLs to URIs?
Yes, please specify. We don't want this to end up in a sort of FUD against
URIs. To the best of my knowledge, it's all really simple, as summarised
Ahem, STD 66 presents a slightly different view. 8-)
The problem is that it's impossible to distinguish identifiers and
locators at the syntax level. It's not possible to uphold the URN/URL
After giving reading STD 66 a quick once over (and it is indeed new -
less than a month old), I see your point. Especially since there is
no way to tell them apart anyway, I think that the proper thing for
2440bis is to call *both* keyserver and policy URLs, URIs instead.
There are a number of reasons, but the main one as I see it is that it
is not our business (as OpenPGP people) to tell people what is
acceptable for their keyserver or policy fields to point to. If
someone wants to point to a book, great. Go ahead. If some
implementation wants to support URNs, also great.
Making this change has no impact on any implementations since URLs are