Re: Patch: Write Amplification Reduction Method (WARM)

On Wed, Apr 5, 2017 at 11:57 PM, Andres Freund <andres(at)anarazel(dot)de> wrote:

> On 2017-04-05 09:36:47 -0400, Robert Haas wrote:
> > By the way, the "Converting WARM chains back to HOT chains" section of
> > README.WARM seems to be out of date. Any chance you could update that
> > to reflect the current state and thinking of the patch?
>
> I propose we move this patch to the next CF. That shouldn't prevent you
> working on it, although focusing on review of patches that still might
> make it wouldn't hurt either.
>
>
Thank you all for the reviews, feedback, tests, criticism. And apologies
for keep pushing it till the last minute even though it was clear to me
quite some time back the patch is not going to make it. But if I'd given
up, it would have never received whatever little attention it got. The only
thing that disappoints me is that the patch was held back on no strong
technical grounds - at least none were clear to me. There were concerns
about on-disk changes etc, but most on-disk changes were known for 7 months
now. Reminds me of HOT development, when it would not receive adequate
feedback for quite many months, probably for very similar reasons - complex
patch, changes on-disk format, risky, even though performance gains were
quite substantial. I was much more hopeful this time because we have many
more experts now as compared to then, but we probably have equally more
amount of complex patches to review/commit.

I understand that we would like this patch to go in very early in the
development cycle. So as Alvaro mentioned elsewhere, we will continue to
work on it so that we can get it in as soon as v11 tree open. We shall soon
submit a revised version, with the list of critical things so that we can
discuss them here and get some useful feedback. I hope everyone understands
that the feature of this kind won't happen without on-disk format changes.
So to be able to address any concerns, we will need specific feedback and
workable suggestions, if any.

Finally, my apologies for not spending enough time reviewing other patches.
I know its critical, and I'll try to improve on that. Congratulations to
all whose work got accepted and many thanks to all reviewers/committers/CF
managers. I know how difficult and thankless that work is.

Thanks,
Pavan

--
Pavan Deolasee http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services