Horton

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

Timur Magomedov <t.magomedov@postgrespro.ru>

View thread
From: Timur Magomedov <t.magomedov@postgrespro.ru>
To: Peter Smith <smithpb2250@gmail.com>
Cc: Tomas Vondra <tomas@vondra.me>, "Aya Iwata (Fujitsu)" <iwata.aya@fujitsu.com>, "pgsql-hackers@lists.postgresql.org" <pgsql-hackers@lists.postgresql.org>
Date: 2025-06-27T14:24:41Z
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 Wed, 2025-06-25 at 12:22 +1000, Peter Smith wrote:
> Hi.
> 
> Here are the latest v9* patches. These have following changes:
> 
> 0001
> -  fixes some of the minor quirks reported by Tomas [1].
> 
> 0002
> - fixes to added address some of Timur's feedback/patches [2]
> - test code updated to remove unsupported type
> - test code updated to remove dynamic costings
> 
Hello Peter!

Thanks for updates.
I was trying to move some code from Postgres core patch to contrib/vci.
Started with moving VCI options from StdRdOptions struct and
reloptions.c, reloptions.h files into contrib/vci like Tomas suggested
earlier. Getting new relopt kind using add_reloption_kind() instead of
hardcoded RELOPT_KIND_VCI and so on, just like in contrib/bloom.

During this work I've found that those VCI-specific options,
"vci_column_ids" and "vci_dropped_column_ids", are only read in case
vci_IsExtendedToMoreThan32Columns() is true. And
vci_IsExtendedToMoreThan32Columns() itself checks if vci_column_ids
option is non-empty string to return true. AFAIK no one sets those
options and vci_IsExtendedToMoreThan32Columns() always returns false.

There are comments in code mentioning that one can create VCI index
with more than 32 columns using vci_create(). However, there is no such
a function anywhere. I guess it is now impossible to create VCI index
with more than 32 (INDEX_MAX_KEYS) columns. Maybe the options were used
to store column information for indexes created with vci_create() but
this function has gone.

So, giving the ideas above, can we remove those vci_column_ids,
vci_dropped_column_ids options and all the code that assumes
vci_IsExtendedToMoreThan32Columns() to be true? This would make core
patch a bit shorter, specifically there will be no changes in
reloptions.{c,h} files and no changes in StdRdOptions struct in rel.h.

-- 
Regards,
Timur Magomedov