bug#24603: [PATCH 0/3] Case table updates

[Eli Zaretskii <eliz@gnu.org>]

> From: Michal Nazarewicz <mina86@mina86.com>
> Cc: eliz@gnu.org
> Date: Tue, 18 Oct 2016 00:03:42 +0200
> 
> As I continue working on the patchset, it keeps on growing.  To
> somewhat limit that, I’ll start applying the patches.

This is up to you to some degree, but I'd like to point out that it is
not necessary to apply patches piecemeal.  You can merge to, or rebase
on, master everything in one go when you are done, there should be no
technical difficulties with that whatsoever.

Another potential issue I'd like us to avoid is to modify the same
parts of the code several times in related commits.  If that happens,
I'd prefer a single commit that changes them only once.

That said, separate pushes are justified if the parts you push provide
self-contained significant features or improvements.

> The first two map to the first two from original sumbission.  The
> first gained tests for byte-8 characters and the second includes
> changes requested by Eli.
> 
> The third is a new patch.
> 
> Michal Nazarewicz (3):
>   Add tests for casefiddle.c
>   Generate upcase and downcase tables from Unicode data
>   Don’t generate ‘X maps to X’ entries in case tables

Can't the 3rd patch break some code which assumes the current state of
affairs, i.e. that the case-table entries for characters with no case
variants are identical to the character itself?  IOW, this sounds like
an incompatible change, so it should be mentioned as such in NEWS, and
perhaps we should make sure we don't break too much code out there,
not sure if that is possible.

What kind of memory savings does this produce, in terms of memory
footprint of a running Emacs process?

I have no objections or comments to the other 2 patches.

Thanks.