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: Andrew Dunstan <andrew@dunslane.net>, jian he <jian.universality@gmail.com>, Erik Rijkers <er@xs4all.nl>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2023-09-27T13:55:04Z
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

On Thu, Sep 21, 2023 at 9:41 PM Amit Langote <amitlangote09@gmail.com> wrote:
> On Thu, Sep 21, 2023 at 5:58 PM Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
> > I keep looking at 0001, and in the struct definition I think putting the
> > escontext at the bottom is not great, because there's a comment a few
> > lines above that says "XXX: following fields only needed during
> > "compilation"), could be thrown away afterwards".  This comment is not
> > strictly true, because innermost_caseval is actually used by
> > array_map(); yet it seems that ->escontext should appear before that
> > comment.
>
> Hmm.   Actually, we can make it so that *escontext* is only needed
> during ExecInitExprRec() and never after that.  I've done that in the
> attached updated patch, where you can see that ExprState.escontext is
> only ever touched in execExpr.c.   Also, I noticed that I had
> forgotten to extract one more expression node type's conversion to use
> soft errors from the main patch (0003).  That is CoerceToDomain, which
> I've now moved into 0001.
>
> > Also, ->escontext's own comment in ExprState seems to be saying too much
> > and not saying enough.  I would reword it as "For expression nodes that
> > support soft errors.  NULL if caller wants them thrown instead".  The
> > shortest I could make so that it fits in a single is "For nodes that can
> > error softly. NULL if caller wants them thrown", or "For
> > soft-error-enabled nodes.  NULL if caller wants errors thrown".  Not
> > sure if those are good enough, or just make the comment the whole four
> > lines ...
>
> How about:
>
> +   /*
> +    * For expression nodes that support soft errors.  Set to NULL before
> +    * calling ExecInitExprRec() if the caller wants errors thrown.
> +    */

Maybe the following is better:

+   /*
+    * For expression nodes that support soft errors.  Should be set to NULL
+    * before calling ExecInitExprRec() if the caller wants errors thrown.
+    */

...as in the attached.

Alvaro, do you think your concern regarding escontext not being in the
right spot in the ExprState struct is addressed?  It doesn't seem very
critical to me to place it in the struct's 1st cacheline, because
escontext is not accessed in performance critical paths such as during
expression evaluation, especially with the latest version.  (It would
get accessed during evaluation with previous versions.)

If so, I'd like to move ahead with committing it.  0002 seems almost there too.

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