[
Lists Home |
Date Index |
Thread Index
]
> From: Emmanuil
> Sorry, for being negative in this post, but I'm sure that many
> (including me) are against any form of central registries.
I expected negativity. The flames keep me toasty warm.
> But even beyond that, the word registry (central or not - a
> distributed
> registry is achievable)- means more request>response lag.
Nope. You don't need to look up the prefix in a registry. The prefix is
just a short namespace identifier, one which can be assigned a semantic
meaning, but that's up to whoever registered the prefix.
> This
> complicates the design further as a large part of use-cases
> will demand
> a local mirror of the registry.
Nope, again. Iff you want to attach semantics, then some mirror may be
needed. There are plenty of cases already where a namespace id is a real
URI to a real resource on the web. But it doesn't mean anybody's actually
looking at it.
> People tired of typing URIs use entities (a common thing among RDF
> people). Do we want prefixes to break out of the document scope?
Nothing is breaking out of the document scope, except the authority to
create a prefix designated as 'registered'. Provisional prefixes could use
the same namepaces in xml mechanisms as they exist now, but I confess that I
haven't thought through how you'd mix provisional and registred namespaces
in the same document. One could designate a different separator character
for registered prefixes, but the one good one has already been taken.
>
> In general I agree that a new namespaces mechanism would be a
> good thing
> (if designed utilising community feedback) but the problem is
> very much
> connected with the (miss) use of URIs in general. The current
> situation
> is just a pot where URIs as vocabularies, namespaces, locators and
> identifiers boil all together. Different people have different views
> about what the dish will taste like.
Notice that I said that a URI is optional for a registered prefix!
|
