Re: Declarative partitioning

On Thu, Feb 18, 2016 at 12:41 AM, Amit Langote <
Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp> wrote:

> START [ EXCL ] (startval) END [ INCL ] (endval)
>
> That is, in range type notation, '[startval, endval)' is the default
> behavior. So for each partition, there is at least the following pieces of
> metadata:
>

This is really close, and if it is what we ended up with we would be able
to use it.

I suggest that the range notation can be used even when no suitable range
type exists.

I assume the code for parsing a range spec regardless of data type already
exists, but in case it doesn't, take a range spec of unknown type:

[x,y)

x and y are either going to be raw strings or doublequoted strings with
possible doublequote escapes, each of which would be coercible into the the
type of the partition column.

In other words, if your string values were 'blah , blah ' and 'fabizzle',
the [) range spec would be ["blah , blah ",fabizzle).

Using regular range specs syntax also allows for the range to be unbounded
in either or both directions, which is a possibility, especially in newer
tables where the expected distribution of data is unknown.

p.s. Sorry I haven't been able to kick the tires just yet. We have a very
good use case for this, it's just a matter of getting a machine and the
time to devote to it.