Thread: BUG #15566: can't insert money type

The following bug has been logged on the website:

Bug reference:      15566
Logged by:          Yusuke Sonomura
Email address:      yusuke.sonomura@jp.fujitsu.com
PostgreSQL version: 10.3
Operating system:   Solaris
Description:

-Issue
I can't INSERT numeric data to money column on Solaris.
And I can't understand what error message says. I think codeset should be
"UTF-8".

--How to reproduce
export LANG=ja_JP.UTF-8
initdb --encoding="UTF-8" --locale="ja_JP.UTF-8" --lc-collate="C"
--lc-ctype="C" --lc-messages="ja_JP.UTF-8" --lc-monetary="ja_JP.UTF-8"
--lc-numeric="ja_JP.UTF-8" --lc-time="ja_JP.UTF-8" -D inst1
pg_ctl start -D inst1 
psql -d postgres<<EOF
CREATE TABLE test (a money);
INSERT INTO test values (2);
EOF

--Output I got
CREATE TABLE
2018-12-21 22:15:41.389 JST [8295] WARNING:  could not determine encoding
for locale "ja_JP.UTF-8": codeset is "646"
WARNING:  could not determine encoding for locale "ja_JP.UTF-8": codeset is
"646"
2018-12-21 22:15:41.390 JST [8295] ERROR:  default conversion function for
encoding "" to "UTF8" does not exist
2018-12-21 22:15:41.390 JST [8295] STATEMENT:  INSERT INTO test values
(2);
ERROR:  default conversion function for encoding "" to "UTF8" does not
exist

--Output I expected
Just success of insertion.

--Installation I used
I got postgresql-10.3-S11.sparc-64.tar.bz2 from 
https://www.postgresql.org/ftp/binary/v10.3/solaris/solaris11/sparc/

--Version
10.3

--Platform information
Solaris 11
SunOS s023g023 5.11 11.3 sun4v sparc sun4v

--my environment
Memory size: 4096 Megabytes
CPU 1649MHz*2

--my output of locale -a
C
POSIX
de_DE.UTF-8
en_US.UTF-8
es_ES.UTF-8
fr_FR.UTF-8
it_IT.UTF-8
ja_JP.UTF-8
ja_JP.eucJP
ko_KR.UTF-8
pt_BR.UTF-8
zh_CN.UTF-8
zh_TW.UTF-8

=?utf-8?q?PG_Bug_reporting_form?= <noreply@postgresql.org> writes:
> -Issue
> I can't INSERT numeric data to money column on Solaris.
> And I can't understand what error message says. I think codeset should be
> "UTF-8".

> --How to reproduce
> export LANG=ja_JP.UTF-8
> initdb --encoding="UTF-8" --locale="ja_JP.UTF-8" --lc-collate="C"
> --lc-ctype="C" --lc-messages="ja_JP.UTF-8" --lc-monetary="ja_JP.UTF-8"
> --lc-numeric="ja_JP.UTF-8" --lc-time="ja_JP.UTF-8" -D inst1

Hm.  I think you're shooting yourself in the foot by specifying
--lc-ctype="C"; in the locale world, that setting is what determines
the character encoding.  If you want to use UTF8 then LC_CTYPE had
better match that.

This combination of LC_xxx settings does seem to work more or less sanely
on glibc, but it doesn't surprise me that it doesn't work on Solaris.

> --Output I got
> 2018-12-21 22:15:41.389 JST [8295] WARNING:  could not determine encoding
> for locale "ja_JP.UTF-8": codeset is "646"
> WARNING:  could not determine encoding for locale "ja_JP.UTF-8": codeset is
> "646"

I suppose "646" here refers to ISO 646, which is basically ASCII.
Maybe we should recognize codeset "646" as a synonym for PG_SQL_ASCII.
But I have a feeling that that'd just leave you with some other, even
less decipherable failure later, given this combination of LC_xxx
settings.

            regards, tom lane