On Wed, May 27, 2026 at 09:54:16AM +0200, Patrice Dumas wrote: > On Tue, May 26, 2026 at 06:19:56PM +0100, Gavin Smith wrote: > > I think that we should not uppercase the index entries before getting the > > collation key, except in the case of USE_UNICODE_COLLATION=0. > > It has a performance cost (in C), which is visible now that getting the > collation string is fast. > > > When using USE_UNICODE_COLLATION=0, it is fine to carry on doing it the way > > we are currently doing it - uppercase before getting the sort keys. There > > is no need for any extra complications to always get upper and lower case in > > exactly the right order. > > What about not uppercasing in that case either? Setting > USE_UNICODE_COLLATION=0 is likely to be rare, and choosen for speed in > cases where the sort order is not so important.
I'm happy with that, as it's probably fairly rare for users to set USE_UNICODE_COLLATION=0, and it would be less necessary to speed up the program.
