Re: pgsql: Make parser rely more heavily on the ParseNamespaceItemdata str

On Thu, Jan 2, 2020 at 6:41 PM Tom Lane <tgl@sss.pgh.pa.us> wrote:
>
> Tom Lane <tgl@sss.pgh.pa.us> writes:
> > Make parser rely more heavily on the ParseNamespaceItem data structure.
>
> Hmm, the buildfarm seems to think this screwed up something affecting
> collations.  Looking ...

My animal is using fr_FR collation by default, and
https://buildfarm.postgresql.org/cgi-bin/show_log.pl?nm=lapwing&dt=2020-01-02%2016%3A40%3A21
is reporting:

diff -U3 /home/postgres/buildroot/HEAD/pgsql.build/src/test/regress/expected/collate.out
/home/postgres/buildroot/HEAD/pgsql.build/src/test/regress/results/collate.out
--- /home/postgres/buildroot/HEAD/pgsql.build/src/test/regress/expected/collate.out
2019-06-04 18:32:50.356908034 +0000
+++ /home/postgres/buildroot/HEAD/pgsql.build/src/test/regress/results/collate.out
2020-01-02 17:21:27.696908000 +0000
@@ -540,18 +540,18 @@
 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM
collate_test1)) ORDER BY 1;
  unnest
 --------
- ABD
- Abc
  abc
+ Abc
+ ABD
  bbc
 (4 rows)

 SELECT * FROM unnest((SELECT array_agg(b ORDER BY b) FROM
collate_test2)) ORDER BY 1;
  unnest
 --------
- ABD
- Abc
  abc
+ Abc
+ ABD
  bbc
 (4 rows)

Is ORDER BY 1 COLLATE "C" an option here?