Half an year on Alpine: just musl aside ~ @jutty
d@nny disc@ mc²
boosted
Just designed and tested an algorithm to build a reverse-iterator on top of an iterator that can only run in forwards direction (part of the #musl collation project) and it seems to be good!
63 underlying forward-iterate steps to perform 21 reverse-iterate steps in simple test case.
Just designed and tested an algorithm to build a reverse-iterator on top of an iterator that can only run in forwards direction (part of the #musl collation project) and it seems to be good!
63 underlying forward-iterate steps to perform 21 reverse-iterate steps in simple test case.
Basically, #musl's locale support has been pretty abysmal for a long time.
This is in contrast to multilingualization, which has always been first-class. M17N basically means ability to represent data in your own and other languages concurrently, and is about character set and avoiding artificial limitations. Locale/localization, OTOH, means the ability of the software to adjust its behavior and presentation to the norms of a particular cultural context, and that's what we've failed at.
About 10 years ago, I wrote the initial locale implementation, but it never received testing/feedback, and had core data representation bugs that basically made it not usable.
5/N
This time around, in addition to funding, we have folks from @postmarketOS on the project to give feedback on integration and actually meeting the needs for localized UI in end-user-facing applications.
A decade ago, #musl was pretty much only used in embedded and container spaces, and occasionally desktops of power users mostly sticking to a non-localized/English-only interface.
A lot has changed since then.
6/N
The big exciting news though is that some work I've been wanting to do on #musl for a long time now has sponsorship, thanks to a grant from NLnet and NGI Zero Core.
This was previously announced on the musl mailing list, but I'd like to go into some more detail here.
4/N
Basically, #musl's locale support has been pretty abysmal for a long time.
This is in contrast to multilingualization, which has always been first-class. M17N basically means ability to represent data in your own and other languages concurrently, and is about character set and avoiding artificial limitations. Locale/localization, OTOH, means the ability of the software to adjust its behavior and presentation to the norms of a particular cultural context, and that's what we've failed at.
About 10 years ago, I wrote the initial locale implementation, but it never received testing/feedback, and had core data representation bugs that basically made it not usable.
5/N
First, yes it's been a long time since a release. I was out of my country under rather extraordinary circumstances for 3 months basically from the time of the electoral part of the fascist coup to when the regime was just about to start implementing things, and it's taken me a while to get back to any kind of routine and catch up on the backlog of things I'd forgotten need attention. But that's all going well, and I'm aiming for a release somewhere in the next 10-25 days.
3/N
The big exciting news though is that some work I've been wanting to do on #musl for a long time now has sponsorship, thanks to a grant from NLnet and NGI Zero Core.
This was previously announced on the musl mailing list, but I'd like to go into some more detail here.
4/N