Re: CF entries for 17 to be reviewed

> On 6 Mar 2024, at 18:49, Andrey M. Borodin <x4mmm(at)yandex-team(dot)ru> wrote:
>
> Here are statuses for "Refactoring" section:

I've made a pass through "Replication and Recovery" and "SQL commands" sections.
"SQL commands" section seems to me most interesting stuff on the commitfest (but I'm far from inspecting every thing thread yet). But reviewing patches from this section is not a work for one CF definitely. Be prepared that this is a long run, with many iterations and occasional architectural pivots. Typically reviewers work on these items for much more than one month.

Replication and Recovery
* Synchronizing slots from primary to standby
Titanic work. A lot of stuff already pushed, v108 is now in the discussion. ISTM that entry barrier of jumping into discussion with something useful is really high.
* CREATE SUBSCRIPTION ... SERVER
The discussion stopped in January. Authors posted new version recently.
* speed up a logical replication setup (pg_createsubscriber)
Newest version posted recently, but it fails CFbot's tests. Pinged authors.
* Have pg_basebackup write "dbname" in "primary_conninfo"?
Some feedback and descussin provided. Switched to WoA.

SQL Commands
* Add SPLIT PARTITION/MERGE PARTITIONS commands
Cool new commands, very useful for sharding. CF item was RfC recently, need review update after rebase.
* Add CANONICAL option to xmlserialize
Vignesh C and Chapman Flack provided some feedback back in October 2023, but the patch still needs review.
* Incremental View Maintenance
This is a super cool feature. IMO at this point is too late for 17, but you should consider reviewing this not because it's easy, but because it's hard. It's real rocket science. Fine-grained 11-step patchset, which can change a lot of workloads if committed. CFbot finds some failures, but this should not stop anyone frome reviewing in this case.
* Implement row pattern recognition feature
SQL:2016 feature, carefully split into 8 steps. Needs review, probably review in a long run. The feature seems big. 3 reviewers are working on this, but no recent input for some months.
* COPY TO json
Active thread with many different authors proposing different patches. I could not summarize it, asked CF entry author for help.
* Add new error_action COPY ON_ERROR "log"
There's an active discussion in the thread.
* add COPY option RECECT_LIMIT
While the patch seems much similar to previous, there's no discussion in this thread...

Thanks!

Best regards, Andrey Borodin.