Re: pgsql: Implement waiting for given lsn at transaction start

On Wed, Apr 8, 2020 at 5:18 AM Michael Paquier <michael(at)paquier(dot)xyz> wrote:
> On Wed, Apr 08, 2020 at 06:07:23AM +0530, Amit Kapila wrote:
> > I think we have two options now (a) Provide feedback on the thread for
> > syntax and see what best we can do in that regard (b) Revert and try
> > it for PG14. I think generally building consensus on syntax at this
> > stage is difficult but we can try if we want this feature for this
> > release. I am not very happy that it went in without more discussion
> > but OTOH, this is not a very big feature and if we agree on syntax
> > this can be part of PG13. I think code also needs some more review.
>
> I have not followed this stuff as much as I would have liked, but my
> overall impression of the patch is that the disagreements around the
> grammar syntax with the addition of three keywords which I guess are
> heavily used in AS aliases because they are generic terms (including
> lsn!), and the issues mentioned on the thread point out that this
> patch was not ready to be merged, and that this has been rushed as per
> the feature freeze deadline. Post feature-freeze is not the time to
> discuss feature redesign, so I think that this should be reverted, and
> proposed again for 14.

In my original idea, we can assume simplified syntax to be agreed.
And other issues could be easily resolved if arise. But it appears we
have serious complains to even simplified syntax. For sure, we
shouldn't discuss it post feature freeze. I've reverted this commit.

------
Alexander Korotkov
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company