Re: [hackers] [lchat] use libgrapheme instead of libutf || Jan Klemkow

From: Laslo Hunhold <dev_AT_frign.de>
Date: Sun, 2 Oct 2022 02:37:12 +0200

On Sun, 2 Oct 2022 02:01:34 +0200 (CEST)
git_AT_suckless.org wrote:

Dear Jan,

> commit dbc8751dc6c034967d2b3133a58a627834992e8c
> Author: Jan Klemkow <j.klemkow_AT_wemelug.de>
> AuthorDate: Sun Oct 2 00:59:19 2022 +0200
> Commit: Jan Klemkow <j.klemkow_AT_wemelug.de>
> CommitDate: Sun Oct 2 01:00:03 2022 +0200
>
> use libgrapheme instead of libutf

thanks for putting forward the trust and using libgrapheme for your
application!

I am currently in the process of heavy refactorization in preparation
of version 2 (I want to put the code on much more formally-verifiable
fotting), but version 1 is generally stable and there are no known
bugs.

Until now, I refactored the case-, character- and line-functions and
they are working perfectly, which the unit-tests reflect. The word- and
sentence-functions have more complex state-handling that requires me to
think of a fitting data-structure, and I know of some edge-cases where
they might fail (e.g. NUL-terminated strings) given the iffy
index-jiggling.

But given you're only using character-break-checks, you're safe. There
will however be an API-change with version 2 where

        grapheme_next_character_break()

is renamed to

        grapheme_next_character_break_utf8().

I know that such changes are always a bad thing and I gave it a lot of
thought, but it's better to change now, where only very few projects
use the library, instead of having to carry this as legacy cruft into
the future.

With best regards

Laslo
Received on Sun Oct 02 2022 - 02:37:12 CEST