Benjamin Good wrote:
> Thanks, working much better now.
>
> For queries like this, how are the results ordered?
>
> For example, if I wanted to page through a list of 2000 unique
> bookmarks, 100 at a time, would I be safe to issue a series of requests
> like:
> http://www.connotea.org/data/bookmarks?num=100&start=0
> http://www.connotea.org/data/bookmarks?num=100&start=100
> ..
>
> is there any chance the resultsets would overlap? Is there a minimum
> interval between requests that would guarantee that the numbering
> remains consistent?
The results are ordered by bookmark creation time - the timestamp that
the first user submitted that particular URL.
You do have a shift problem that arises from paging through a changing
dataset. I have not seen a good solution to this in any API on the web
to date. I thought about it recently, speculating that we might solve
that problem, but hadn't got around to it. I think you could solve the
stepping over of running additions it with an additional parameter for
initial timestamp or id, but there is also the problem of running deletions.
Martin
|
