HI there,
I've committed a tiny bit of code to the postgresql_measures_types project on github:
It faintly shows the direction I'm going in. Over the next few weeks of work in the evenings I'm going to tackle the to-do list:
- Write a generic _distinct functional on the representation of simple measurable functions that sieves the point-wise distinct values (tomorrow?)
- Write type guarding wrappers for the generic point-wise functional (next week?)
- Write aggregate wrappers around the type wrappers: first concatenate the array representations of the simple measurable functions and then passes the array through the type wrappers (next week and two weeks?)
- Write operator wrappers around the type wrappers: concatenate a pair of representations and the pass to the type wrappers (third week?)
- At some point try to find an efficient strategy for a functional that finds the point-wise array and string aggregates, this requires an arbitrary access stack (ugh?)
The underlying idea is that a partition of a strictly totally ordered set into disjoint intervals (ranges) implies an underlying linear structure to the data around which the algorithms can be optimized. This work encapsulates and generalizes the types of analytic queries I have had to write most often over the last couple of years.
Thank you,
Aaron
On Thu, Oct 11, 2012 at 12:07 PM, Stirling Newberry
<stirling.newberry@xigenics.net> wrote:
Interesting, could you post the link when you set it up?
BTW typo it is vein. Though of course autocorrect does not always agree with the author and sometimes creates typos for us.
Cheers.
--
Aaron Sheldon
#67 Westedge Townhouses
