Horton

Re: TM format can mix encodings in to_char()

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Juan José Santamaría Flecha <juanjo.santamaria@gmail.com>
Cc: pgsql-hackers@lists.postgresql.org
Date: 2019-04-21T04:28:48Z
Lists: pgsql-hackers

Attachments

I wrote:
> [ fix-encoding-and-error-recovery-in-cache-locale-time.patch ]

On closer inspection, I'm pretty sure either version of this patch
will break things on Windows, because that platform already had code
to convert the result of wcsftime() to the database encoding; we
were adding code to do a second conversion, which will not go well.

The attached revised patch deletes the no-longer-necessary
platform-specific recoding stanza, in favor of having cache_locale_time
know that it's getting UTF8 rather than something else.  I also
updated a bunch of the related comments.

I don't have any way to test this on Windows, so could somebody
do that?  Manually running the Turkish test cases ought to be enough.

			regards, tom lane

Commits

  1. Repair assorted issues in locale data extraction.