Horton

Re: [WIP]Vertical Clustered Index (columnar store extension) - take2

Peter Smith <smithpb2250@gmail.com>

View thread
From: Peter Smith <smithpb2250@gmail.com>
To: Japin Li <japinli@hotmail.com>
Cc: "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>, Tomas Vondra <tomas@vondra.me>, "Aya Iwata (Fujitsu)" <iwata.aya@fujitsu.com>, Timur Magomedov <t.magomedov@postgrespro.ru>, shveta malik <shveta.malik@gmail.com>
Date: 2025-07-23T04:07:59Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Mark ItemPointer arguments as const throughout

  2. Reorganize GUC structs

  3. Eliminate XLOG_HEAP2_VISIBLE from vacuum phase III

  4. Update various forward declarations to use typedef

  5. Pathify RHS unique-ification for semijoin planning

  6. Revert "Don't lock partitions pruned by initial pruning"

  7. For inplace update, send nontransactional invalidations.

  8. Remove nearly-unused SizeOfIptrData macro.

On Tue, Jul 22, 2025 at 8:12 PM Japin Li <japinli@hotmail.com> wrote:
...
> 1.
> I encountered another crash while checking VCI's internal relations.
>
> rm -rf demo
> initdb -D demo
> cat <<EOF >demo/postgresql.auto.conf
> shared_preload_libraries = 'vci'
> max_worker_processes = '20'
> EOF
>
> pg_ctl -D demo start
>
> cat <<EOF | psql postgres
> CREATE EXTENSION vci;
> CREATE TABLE t (id int, info text);
> CREATE INDEX ON t USING vci (id);
> SELECT relname FROM pg_class WHERE relname ~ '^vci_*' LIMIT 1 \gset
> SELECT * FROM :relname;
> \d+ :relname
> REFRESH MATERIALIZED VIEW :relname;
> EOF
>
> VCI's definition of internal relations as materialized views lacks the
> corresponding view bodies.
>
> +    /*
> +     * @see
> +     * https://www.postgresql.jp/document/9.4/html/catalog-pg-rewrite.html
> +     */
> +    new_rel_reltup->relhasrules = true;
> +
> +    new_rel->rd_att->tdtypeid = new_type_oid;
> +
> +    InsertPgClassTuple(pg_class, new_rel, new_oid, (Datum) 0, (Datum) 0);
>
> Given that VCI's internal relations are materialized views, are VCI workers
> responsible for their periodic refreshment?
>
> Or is it by design that users are unable to read the internal relations?
>

IIUC, those VCI internal relations (implemented as materialized views)
are entirely managed by VCI logic. Users are not required to be aware
of them, and they definitely are not meant to tamper with them.

The REFRESH that you attempted should have caused a more graceful error, like:
ERROR:  extension "vci" prohibits this operation on view
"vci_0000016482_00000_d"
So, thanks for reporting that the ERROR failed. Investigating...

======
Kind Regards,
Peter Smith.
Fujitsu Australia