Re: initdb initalization failure for collation "ja_JP"

I wrote:
>> One question that I've got is why the ICU portion refuses to load
>> any entries unless is_encoding_supported_by_icu(GetDatabaseEncoding()).
>> Surely this is completely wrong? I should think that what we load into
>> pg_collation ought to be independent of template1's encoding, the same
>> as it is for libc collations, and the right place to be making a test
>> like that is where somebody attempts to use an ICU collation.

Pursuant to the second part of that: I checked on what happens if you
try to use an ICU collation in a database with a not-supported-by-ICU
encoding. We have to cope with that scenario even with the current
(broken IMO) initdb behavior, because even if template1 has a supported
encoding, it's possible to create another database that doesn't.
It does fail more or less cleanly; you get an "encoding "foo" not
supported by ICU" message at runtime (out of get_encoding_name_for_icu).
But that's quite a bit unlike the behavior for libc collations: with
those, you get an error in collation name lookup, along the lines of

collation "en_DK.utf8" for encoding "SQL_ASCII" does not exist

The attached proposed patch makes the behavior for ICU collations the
same, by dint of injecting the is_encoding_supported_by_icu() check
into collation name lookup.

regards, tom lane