| From: | Peter Eisentraut <peter(dot)eisentraut(at)enterprisedb(dot)com> |
|---|---|
| To: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | JSON_TABLE output collations |
| Date: | 2022-06-07 13:19:01 |
| Message-ID: | 9d75ce67-0121-5050-5bec-bf5009db55ce@enterprisedb.com |
| Views: | Whole Thread | Raw Message | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
The present implementation of JSON_TABLE sets the collation of the
output columns to the default collation if the specified data type is
collatable. Why don't we use the collation of the type directly? This
would make domains with attached collations work correctly.
See attached patch for how to change this. I hacked up a regression
test case to demonstrate this.
| Attachment | Content-Type | Size |
|---|---|---|
| 0001-Fix-collation-of-JSON_TABLE-output-columns.patch | text/plain | 2.1 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Jakub Wartak | 2022-06-07 13:29:21 | RE: effective_io_concurrency and NVMe devices |
| Previous Message | Ian Lawrence Barwick | 2022-06-07 13:08:21 | Re: docs: mention "pg_read_all_stats" in "track_activities" description |