Horton

Thread

Commits

  1. Clean up some code and comments in partbounds.c.

  2. Allow partitionwise joins in more cases.

  1. Minor cleanup of partbounds.c

    Etsuro Fujita <etsuro.fujita@gmail.com> — 2020-09-09T13:37:07Z

    Here is a patch for minor cleanup of the partbounds.c changes made by
commit c8434d64c: 1) removes a useless assignment (in normal builds)
and 2) improves comments a little.

Best regards,
Etsuro Fujita

  2. Re: Minor cleanup of partbounds.c

    Alvaro Herrera <alvherre@2ndquadrant.com> — 2020-09-09T17:05:05Z

    On 2020-Sep-09, Etsuro Fujita wrote:

> Here is a patch for minor cleanup of the partbounds.c changes made by
> commit c8434d64c: 1) removes a useless assignment (in normal builds)

LGTM.

> and 2) improves comments a little.

No objection to changing "bounds" to "range bounds".

I think the point other is to replace the only appearance of "dummy
relation" to better match the extensive use of "dummy partition" in this
file.  The concept of a "dummy relation" is well established in the
planner.  I didn't know if "dummy partition" is itself a concept
(apparently in the newfangled partition-wise join stuff), or just
glorified wording to say "a dummy relation that happens to be a
partition".  Looking at is_dummy_partition, apparently a dummy partition
is either a dummy relation or a partition that doesn't have a
RelOptInfo.  So my conclusion is that this wording is okay to change
too.

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

  3. Re: Minor cleanup of partbounds.c

    Etsuro Fujita <etsuro.fujita@gmail.com> — 2020-09-10T09:10:37Z

    On Thu, Sep 10, 2020 at 2:05 AM Alvaro Herrera <alvherre@2ndquadrant.com> wrote:
> On 2020-Sep-09, Etsuro Fujita wrote:
> > Here is a patch for minor cleanup of the partbounds.c changes made by
> > commit c8434d64c: 1) removes a useless assignment (in normal builds)
>
> LGTM.
>
> > and 2) improves comments a little.
>
> No objection to changing "bounds" to "range bounds".
>
> I think the point other is to replace the only appearance of "dummy
> relation" to better match the extensive use of "dummy partition" in this
> file.  The concept of a "dummy relation" is well established in the
> planner.  I didn't know if "dummy partition" is itself a concept
> (apparently in the newfangled partition-wise join stuff), or just
> glorified wording to say "a dummy relation that happens to be a
> partition".  Looking at is_dummy_partition, apparently a dummy partition
> is either a dummy relation or a partition that doesn't have a
> RelOptInfo.  So my conclusion is that this wording is okay to change
> too.

Cool!

I pushed the patch.  Thanks for reviewing!

Best regards,
Etsuro Fujita