Hi!
On Thu, Mar 24, 2011 at 7:07 PM, Zheng Yang <zhengyang4k@gmail.com> wrote:
> Thanks for the slides! As what's mentioned on one of the slides, the
> datasource of FDW can be virtually everything.
> So if there were a FDW for flickr API, the examples will be something like:
> 1) CREATE FOREIGN TABLE flickr_table (photo_id INT, ownerid INT, photo BLOB,
> text TEXT) SERVER flick_fdw OPTIONS (api_key 'AAAA');
> p.s. the text field is used as the keyword for free text search.
> when query from the table:
Yes, but perhaps we could name the 'text' column something like
'search' or ... someone else could help come up with a better name. :)
> 2) SELECT photo FROM flickr_table WHERE text = 'panda' LIMIT 0, 10;
> This will return top ten photos relevant to 'panda'. Is my understanding
> correct?
Yes, the first ten results.
> I've briefly gone through the slides. Regarding the 6 callbacks, is that
> correct to say that a full table scan will always be performed irregardless
> of the sql statement,
> the FDW is blind to the sql query performed, right?
That's correct.
> And can anyone help explain what a planner is? What does this "provide cost
> estimates to planner" mean? Thanks!
Have a look at this for a quick overview:
http://www.postgresql.org/docs/9.0/static/planner-optimizer.html
-selena
--
http://chesnok.com
