On Sat, 2024-03-02 at 15:02 -0800, Jeff Davis wrote:
> Attached v20.
And here's v22 (I didn't post v21).
I committed Unicode property tables and functions, and the simple case
mapping. I separated out the full case mapping changes (based on
SpecialCasing.txt) into patch 0006.
Not a lot of technical changes, but I cleaned up the remaining patches
and put them into a nicer order with nicer commit messages.
0001: Catalog renaming: colliculocale to colllocale and daticulocale to
datlocale.
0002: Basic builtin collation provider that only supports "C".
0003: C.UTF-8 locale for builtin collation provider and collation
pg_c_utf8.
0004: Inline some UTF-8 functions to improve performance
0005: Add a unicode_strtitle() function and move the implementation for
the builtin provider out of formatting.c.
0006: Add full case mapping support
0007: Add PG_UNICODE_FAST locale for builtin collation provider and
collation pg_unicode_fast. This behaves like the standard says
UCS_BASIC should behave -- sort by code point order but use Unicode
character semantics with full case mapping.
0004 and beyond could use some review. 0004 and 0005 are pretty simple
and non-controversial. 0006 and 0007 are a bit more interesting and
could use some discussion if we want to go ahead with full case mapping
in 17.
Regards,
Jeff Davis
