Horton

Re: remaining sql/json patches

amit <amitlangote09@gmail.com>

View thread
From: Amit Langote <amitlangote09@gmail.com>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>, Andrew Dunstan <andrew@dunslane.net>
Date: 2023-07-14T07:13:17Z
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. SQL/JSON: Various improvements to SQL/JSON query function docs

  2. SQL/JSON: Fix some obsolete comments.

  3. SQL/JSON: Fix issues with DEFAULT .. ON ERROR / EMPTY

  4. JSON_TABLE: Add support for NESTED paths and columns

  5. Fix JsonExpr deparsing to emit QUOTES and WRAPPER correctly

  6. Fix typo introduced in 6185c9737

  7. Add basic JSON_TABLE() functionality

  8. Avoid splitting errmsg string to span multiple lines

  9. Add SQL/JSON query functions

  10. Implement various jsonpath methods

  11. Add soft error handling to some expression nodes

  12. Adjust populate_record_field() to handle errors softly

  13. Refactor code used by jsonpath executor to fetch variables

  14. Test EXPLAIN (FORMAT JSON) ... XMLTABLE

  15. Simplify productions for FORMAT JSON [ ENCODING name ]

  16. Add trailing commas to enum definitions

  17. doc: add missing <returnvalue> and whitespace

  18. Add more SQL/JSON constructor functions

  19. Rename a nonterminal used in SQL/JSON grammar

  20. Some refactoring to export json(b) conversion functions

  21. Don't include CaseTestExpr in JsonValueExpr.formatted_expr

  22. Code review for commit b6e1157e7d

  23. Pass constructName to transformJsonValueExpr()

  24. Unify JSON categorize type API and export for external use

  25. Make some indentation in gram.y consistent

  26. Allow most keywords to be used as column labels without requiring AS.

  27. Reduce size of backend scanner's tables.

  28. Use perfect hashing, instead of binary search, for keyword lookup.

Attachments

Hi Alvaro,

On Fri, Jul 14, 2023 at 1:54 AM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> I looked at your 0001.  My 0001 are some trivial comment cleanups to
> that.

Thanks.

> I scrolled through all of jsonfuncs.c to see if there was a better place
> for the new function than the end of the file.  Man, is that one ugly
> file.  There are almost no comments!  I almost wish you would create a
> new file so that you don't have to put this new function in such bad
> company.  But maybe it'll improve someday, so ... whatever.

I tried to put it somewhere that is not the end of the file, though
anywhere would have looked arbitrary anyway for the reasons you
mention, so I didn't after all.

> In the original code, the functions here being (re)moved do not need to
> return a type output function in a few cases.  This works okay when the
> functions are each contained in a single file (because each function
> knows that the respective datum_to_json/datum_to_jsonb user of the
> returned values won't need the function OID in those other cases); but
> as an exported function, that strange API doesn't seem great.  (It only
> works for 0002 because the only thing that the executor does with these
> cached values is call datum_to_json/b).

Agreed about not tying the new API too closely to datum_to_json[b]'s needs.

>  That seems easy to solve, since
> we can return the hardcoded output function OID in those cases anyway.
> A possible complaint about this is that the OID so returned would be
> untested code, so they might be wrong and we'd never know.  However,
> ISTM it's better to make a promise about always returning a function OID
> and later fixing any bogus function OID if we ever discover that we
> return one, rather than having to document in the function's comment
> that "we only return function OIDs in such and such cases".  So I made
> this change my 0002.

+1

> A similar complaint can be made about which casts we look for.  Right
> now, only an explicit cast to JSON is useful, so that's the only thing
> we do.  But maybe one day a cast to JSONB would become useful if there's
> no cast to JSON for some datatype (in the is_jsonb case only?); and
> maybe another type of cast would be useful.  However, that seems like
> going too much into uncharted territory with no useful use case, so
> let's just not go there for now.  Maybe in the future we can improve
> this aspect of it, if need arises.

Hmm, yes, the note in the nearby comment stresses "to json (not to
jsonb)", though the (historical) reason why is not so clear to me.
I'm inclined to leave that as-is.

I've merged your deltas in the attached 0001 and rebased the other
patches.  In 0002, I have now removed RETURNING support for JSON() and
JSON_SCALAR().


--
Thanks, Amit Langote
EDB: http://www.enterprisedb.com