[musl] Announcement: Locale support overhaul project

[musl] Announcement: Locale support overhaul project

[Rich Felker]

Some exciting news for musl:

Thanks to a grant from NLnet, I'm starting on a project to overhaul
the locale support in musl. This will be in collaboration with folks
from postmarketOS, a project which is well-positioned to provide
testing and feedback by virtue of their using musl in an
end-user-facing setting.

The two big technical parts of this project will be:

(1) developing collation (LC_COLLATE, strcoll/strxfrm interfaces)
support based on the Unicode Collation Algorithm (UCA/UTS#10), and

(2) overhauling the data representation for musl locale files so that
it's possible to build working locales (e.g. without the May/May
issue) and represent LC_NUMERIC/LC_MONETARY properties not presently
supported.

While the groundwork for these things began years ago and there are
already some mailing list threads that partly established desired
directions for locale-system changes, lack of feedback from parties
who would actually be using the resulting functionality really held us
back from getting anywhere. Working together with pmOS is going to be
a big boost there, but as always, this is a project that needs
community feedback and consensus building, and everyone is invited to
take place in the design discussions that will take place during this
process.

More details will be coming as this ramps up, in the form of proposed
data formats and algorithms, likely some of them with choices to be
evaluated for tradeoffs.

At the same time I'll be working to wrap up the current release cycle,
with locale changes to be merged in the next release cycle. Most of
the early work on the locale project however will be out-of-tree, in
the form of design drafts, data formats, and proof of concept code for
the algorithms, so these can and will take place in parallel.