[
Lists Home |
Date Index |
Thread Index
]
Robin Berjon wrote:
> Hi all,
>
> I guess this is still the list for discussion of RDDL (yeah, some of
> us are still into it!), if not I'll happily take this elsewhere.
yes, I used RDDL for some time...though have changed my approach ...read on.
> I'm in the process of writing an RDDL document for a vocabulary that
> has up to now three versions, one of which has two profiles and
> another three, for a total of (at least) six different schemata.
meta meta meta data.....
> Since it's a well-behaved XML vocabulary, it doesn't change namespaces
> between minor versions and profiles, and therefore there will be only
> one RDDL document. Of course the latest and greatest schema for the
> largest profile (they are all strict subsets/supersets compared to one
> another) will validate all profiles of all versions, but some people
> will want to validate at one given profile of a given version.
so I guess you apply new namespaces with major versions e.g. whenever u
want to imply no backwards compatibility, whenever this occurs I tend to
just make a seperate RDDL document as well.
> Eris forbid, I'm not trying to open up a discussion on versioning. I
> just want to know if anyone has given thought to distinguishing
> between variants of a vocabulary in an RDDL document. I'm know it's
> pretty much an idle exercise since writing it for humans will likely
> be sufficient, but hey in my extremely copious spare time I was
> thinking that if someone had given some thought to this, or had an
> idea of how to represent it neatly, I'd be happy to pick it up and run
> with it.
I think in the scenario you present you want to have a simple versioning
for consumption by end users, not to track edits, or for packaging
(builds)...
I like the simplicity of RDDL and used it for some time, though just
recently I converted my RDDL documents to ATOM....it felt better to get
away from xlink...and I get things like id, updated, published, rights,
etc xml elements for free, no defaulting to something like Dublin
core....and there are plenty of API's out there to process.
I also like ATOM because of its explicit stipulation to cont processing
when it encounters foreign markup...this to me is important for future uses.
Informing your developer users of a new iteration becomes part of a
standard notification mechanism...its also easy to make an Atom feed
look like a RDDL html document (e.g. std css/xslt transform) as well.
gl, Jim Fuller
ps: when u coming back to Prague?
|
