Re: [18] Unintentional behavior change in commit e9931bfb75 - Mailing list pgsql-hackers

From Jeff Davis
Subject Re: [18] Unintentional behavior change in commit e9931bfb75
Date
Msg-id 5393ccd73d99ea80f03a6477dbacc476483da979.camel@j-davis.com
Whole thread Raw
In response to Re: [18] Unintentional behavior change in commit e9931bfb75  (Noah Misch <noah@leadboat.com>)
Responses Re: [18] Unintentional behavior change in commit e9931bfb75
List pgsql-hackers
On Mon, 2025-04-14 at 13:44 -0700, Noah Misch wrote:
> v18 regc_pg_locale.c uses only the locale_t-extended forms, and it
> aims not to
> depend on LC_CTYPE.  v17 used both tolower() and tolower_l(), but v18
> uses the
> latter only.

Fixed in v2-0001 by rewording the comment slightly.

>
> However, I think v17's
> concept of separate PG_REGEX_ symbols for the default-locale case is
> still the
> right thing for v18.  In other words, this code should change to look
> more
> like v17, with the removal of non-locale_t calls being the main
> change.

I tried that in v2-0003, but I think it ended up worse. Most
pg_wc_xyz() functions don't care if it's the default collation or not,
so there are a lot of duplicate cases.

The previous approach is still there as v2-0002.

Regrads,
    Jeff Davis


Attachment

pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: pg_dump --if-exists --clean when drop index that is partition of a partitioned index
Next
From: Tom Lane
Date:
Subject: Re: Built-in Raft replication