On 29 Jan 2009, at 07:46, Leslie Carr wrote:
> On 28 Jan 2009, at 18:52, Scott Wilson wrote:
>
>> Exactly - Slideshare and YouTube are both "repositories". There is
>> no "repository abstraction layer".
> I was referring to the repository abstraction layer as the layer of
> software that turns a naked file store and a database into a
> coherent repository.
>
>> It would not be meaningful to make one (beyond the level of
>> aggregating their RSS feeds, which I don't think really counts as
>> creating a repository!).
> Surely the fact that YouTube, SlideShare and Scribd are three
> separate but very similar services for video, slides and word
> processor documents seems to make them a very good candidate for
> some kind of aggregation!
Depends on what you think you are looking at. I don't think they are
"separate but very similar", I think they are completely different
from the ground up. The level of similarity is quite trivial - they
run on computers, they "manage stuff", and like pretty much any
website they can offer RSS feeds.
Its not as though these are all Fedora instances with a few
cusotmisations. For example, Youtube runs over a proprietary CDN
backbone optimised for streaming. Slideshare and Scribd are both
custom-built Ruby apps running off Amazon WS.
>> I think one of the big, big fundamental problems in repository-land
>> (both scholarly and learning resources) stems from what I think is
>> a mistaken belief that the "media doesn't matter" and that you
>> *can* have an abstract "stuff" repository. Talk about making a rod
>> for your own back! ;-)
>
> I would agree that when you reduce everything to just
> undistinguished "stuff" then you have done everyone a dis-service.
> But if you can provide media- and task-specific services for a broad
> range of material, surely you are on to a winner?
See above - the most successful repositories are built from the ground
up to provide media and task-specific services for one content type.
They aren't specializations of a core abstraction layer. Therefore,
the value of such an abstraction should be questioned.
> Can you give an example of where you think that things have gone
> wrong?
Take the amount of R&D investment made so far in generic repository
architecture, metadata, and complex objects, and put it on one side of
the scale. Now put realised value into the other side.
>
> --
> Les
|