Re: Foreign Data Wrapper

On Fri, Nov 13, 2015 at 1:46 PM, Big Mike <newoculus(at)gmail(dot)com> wrote:

> Writing a Foreign Data Wrapper and interested in isolating the WHERE
> clause to speed up the access of an indexed file on my filesystem. I'm
> attempting to understand the inner workings of how the data is retrieved so
> I'm writing code to just handle one case at the moment: WHERE clause on a
> single column in the foreign 'table'.
>
> SELECT * FROM t WHERE testval = 1
>
> I have this code so far, an implementation of the IterateForeignScan
> interface.
>
> static TupleTableSlot *
> bIterateForeignScan(ForeignScanState *node) {
> ...
> RestrictInfo *rinfo = (RestrictInfo *)node->ss.ps.qual;
> ...
> }
>
> yet am not familiar with what I need to do to pick apart RestrictInfo in
> order to gather 'testvar', '=', and '1' separately so I can interpret and
> pass those through to my file parser.
>
> Am I going about this the correct way or is there another path I should
> follow?
>

I would look at http://multicorn.org/ which gives you a working python
framework. You subclass their ForeignDataWrapper class, override the
__init__() and execute() functions, and that's about it. The execute()
function has a list called quals that would set you up for the filtering
you want to do.

I would get the foreign data wrapper fully debugged this way before
attempting to refactor in C. And when you do finally re-code, you can see
what multicorn itself did to implement your filters.