With a relational streaming database, you can perform relational (SQL)
queries on the public timeline or the datamining feed and get real-
time results. This isn't exactly what you are looking for, but it
might meet your needs.
I know the folks at www.sqlstream.com have set this up. I'm sure that
other vendors such as Coral8 and Streambase could also do the same.
-John
On Mar 22, 1:28 pm, Zac Bowling <zbowl...@gmail.com> wrote:
> I would love it if Twitter would develop an equivalent to Facebook's
> FQL, Yahoo's YQL, Amazon's SimpleDB, or Google's GQL (used for app
> engine data storage).
>
> Basically an abstracted SQL-like query engine for doing queries and
> getting back data the data you want using virtual tables of different
> data twitter serves up.
>
> You could do something basic like:
>
> SELECT StatusID, UserID, Text FROM StatusUpdates as S
> WHERE
> S.UserID in (SELECT UserID FROM SocialGraph WHERE FollowerUseringID
> = MYUSERID) and
> S.StatusID > LASTID
> ORDER BY S.StatusID DESC
> LIMIT 200
>
> to get a basic user's following timeline or whatever. From there you
> can build on from that and get a bit more complex.
>
> It could even build on from just query syntax to modify and destructive calls.
>
> Maybe something like:
> DELETE FROM StatusUpdates WHERE StatusID = 200102;
>
> or:
> INSERT INTO StatusUpdates(text,replyToStatusID,replyToUserID) VALUES
> ('@johnsmith hello',123601020,235133);
>
> or:
> UPDATE StatusUpdates SET favorite = TRUE WHERE StatusID = 123601020;
>
> You could do it where you do an HTTP get/post with a query like above
> to twitter's rest api, and the results could come back as JSON or XML
> or whatever.
>
> Some concepts like this could be done in a local side wrapper (like
> I've seen a SQL bridge for MSSQL for twitter on here a while back) but
> it would be awesome if these were processed twitter server side. If
> done right, it can save on overhead on both twitter and from the
> client side.
>
> Like in one case I have where I'm hitting the following timeline, I'm
> missing something out of the user structure that you get back from
> that, so I turn around and do another user call on user for each tweet
> to get that data. Half the data I get back in both cases don't use on
> both calls but it would be awesome to be able to get that data in one
> call.
>
> A lot to consider around optimization and limits and a bit of work to
> build it but I think something like that would be really useful.
>
> Zac
