Edd Dumbill <edd@...> wrote:

> This is a good idea.

Thanks Edd. But you can snap out of journalism mode now. :-) Let's get some
active voice going! (Sorry, I just spent my whole morning reading the "news"
and I'm getting a little tired of the journalistic tone.)

> * First, we recognize that not every content producer has this concept,
> which means it probably ought to be an optional property. I'm undecided
> that it should be a module or not. I think it may be better to have it
> in the core yet mark its as optional.

I agree, but I wasn't sure exactly what the distinction between core and
modules were. If possible, I'd love for this to be core.

> I recommend furthermore the use of
> URNs (though see NOTE below) according to the scheme:
>
> urn:domainID:thread:idstring

I assume that the change from dots and slashes to dashes is required by the
URN scheme, but if we're going to base our domainIDs on URLs anyway,
wouldn't it be better to just use URLs? Perhaps something like:

rss://xmlhack.com/499

or

http://xmlhack.com/rss-thread/499

which could even be a PURL redirecting to the current location of the story.

> NOTE: Although I recommend URNs, any URI should be just fine, but not
> having something resolvable reinforces the 'opaque unique ID' aspect of
> this property. It's the nearest to the flexibility of PCDATA while
> observing the need for universal uniqueness.

Hmm, I'd like to have something resolvable, even if it's an opaque ID.
Perhaps O'Reilly could sponsor a server, in association with Meerkat, that
will allow people to stake out a reserved namespace to place these IDs in.
Then, the URL could point to a list of information collected about the
thread, tracking it through various aggregators, feeds, etc. and collecting
all the data it could on it. This server could also be used as a database of
available RSS feeds, as well as maybe keeping some history of them, and a
good source for information like the RSS surveys.

Of course, it would be nice if this was somehow separate from O'Reilly
(which is why I didn't suggest Meerkat as the host), to emphasize the
non-commercial nature of the project. If no one else is interested, perhaps
I could set such a server up. Then the addresses would look like:

http://rss.example.com/xml-hack/499

Where xml-hack would be registered ahead of time, to reserve the namespace.
Then we could also assemble feeds based on the various taxonomies, etc. A
lot of possibilities.

> * Thirdly, we need to pick a name.

Agreed, but I don't know what to call it. None of your suggestions really
jump out at me.

--
Aaron Swartz |"This information is top security.
<http://swartzfam.com/aaron/>| When you have read it, destroy yourself."
<http://www.theinfo.org/> | - Marshall McLuhan