On 05.06.25 22:47, Jeff Davis wrote:
While we're at it, CTYPE is not very descriptive for a user-facing
name. And COLLATE has become overloaded (expression clause,
pg_collation object, ordering, or the superset of behaviors that
includes CTYPE). Let's consider more user-friendly naming for the
markers:
CASE: lower/upper/initcap/fold behavior
CLASS: char classifications such as [[:punct:]]
ORDER: comparisons
Internally, at least for the foreseeable future, CASE and CLASS would
be the same. They'd just be different markers to record the user's
intent.
Under what scenario would they become different, and how would that
matter in practice?
I would be worried that this could confuse users and they would apply
these incorrectly, if the differences are too fine.
