Re: SQL:2023 JSON simplified accessor support

Alexandra Wang <alexandra.wang.oss@gmail.com>

From: Alexandra Wang <alexandra.wang.oss@gmail.com>

To: Peter Eisentraut <peter@eisentraut.org>

Cc: Andrew Dunstan <andrew@dunslane.net>, Nikita Glukhov <glukhov.n.a@gmail.com>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>, "David E. Wheeler" <david@justatheory.com>, jian he <jian.universality@gmail.com>, matheusssilv97@gmail.com

Date: 2025-02-27T23:23:30Z

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 →

Add test coverage for indirection transformation
- 64492917280a 19 (unreleased) landed
Fix typo in comment
- 81a61fde84ff 19 (unreleased) landed
Implementation of subscripting for jsonb
- 676887a3b0b8 14.0 cited

Attachments

v9-0003-Export-jsonPathFromParseResult.patch (application/octet-stream) patch v9-0003
v9-0002-Allow-Generic-Type-Subscripting-to-Accept-Dot-Not.patch (application/octet-stream) patch v9-0002
v9-0004-Extract-coerce_jsonpath_subscript.patch (application/octet-stream) patch v9-0004
v9-0001-Allow-transformation-of-only-a-sublist-of-subscri.patch (application/octet-stream) patch v9-0001
v9-0006-Implement-read-only-dot-notation-for-jsonb.patch (application/octet-stream) patch v9-0006
v9-0007-Allow-wild-card-member-access-for-jsonb.patch (application/octet-stream) patch v9-0007
v9-0005-Eanble-String-node-as-field-accessors-in-generic-.patch (application/octet-stream) patch v9-0005

Hello hackers,

On Thu, Feb 27, 2025 at 9:46 AM Alexandra Wang <alexandra.wang.oss@gmail.com>
wrote:

> Summary of changes:
>
> v8-0001 through v8-0005:
> Refactoring and preparatory steps for the actual implementation.
>
> v8-0006 (Implement read-only dot notation for JSONB):
> I removed the vars field (introduced in v7) from JsonbSubWorkspace
> after realizing that JsonPathVariable is not actually needed for
> dot-notation.
>
> v8-0007 (Allow wildcard member access for JSONB):
> I'm aware that the #if 0 in check_indirection() is not ideal. I
> haven’t removed it yet because I’m still reviewing other cases—beyond
> our JSONB simplified accessor use case—where this check should remain
> strict. I’ll post an additional patch to address this.
>

I made the following minor changes in v9:
- More detailed commit messages
- Additional tests
- Use "?column?" as the column name for trailing .*.

Other than that, the patches remain the same as the previous
version:
0001 - 0005: preparation steps for the actual implementation and do
not change or add new behavior.
0006: jsonb dot notation and sliced subscripting
0007: jsonb wildcard member access

Thanks,
Alex