Hi list, I'd propose the following steps - open for discussion and as you might notice more vague towards the end: preparations: - choose a name (my suggestion: archlinux32) - set up / use some git hosting (e.g. github) for maintenance of build-system, website, package-source-trees (there are two from archlinux.org: svntogit/community.git and svntogit/packages.git?), possibly our own packages (e.g. archlinux32-keyring?) - clone, understand and modify archlinuxarm's plugbuild (does this have some sort of documentation? - sry, I'm quite new to perl) testing: - set up build-clients (do they need to run i486/i586/i686 or can we cross compile from x86_64?) - compile, install, test packages - keep this up in parallel to official archlinux.org's i686 (so we have a sane benchmark and/or fallback) production: - hopefully change nothing but references in /etc/pacman.d/mirrorlist on production systems ;-) As Bartłomiej pointed out, it would be nice to have i486 and i586 supported, too (*yay*, I finally can run arch on my router!). This shouldn't be a big deal if we really use archlinuxarm's plugbuild - they support compiling for multiple different architectures anyway. One thing I have no clue of, is how the isos are created. But probably it's straight forward if we have some system running on archlinux32 and installed "archiso". regards, Erich On 30.01.2017 23:42, Bartłomiej Piotrowski wrote:
Hi everyone,
sorry for writing so late but there is limit of things I can accomplish every day and that list always includes some cat pictures.
As you are most likely aware, I recently announced the deprecation of i686 support. However I would like this to turn out into a second tier architecture.
I can imagine that there is no need for huge team of packagers; as i686 isn't very different from x86_64, it will be mostly "catch up" game against base distribution (Arch). ARM team already did this work[1] so it's mostly about setting it up with few build servers. ARM port also maintains a separate git repository with modified PKGBUILDs that wouldn't otherwise build on ARM. I'd probably consider targeting i486 or i586 instead to include more old hardware. i686 misses the fanciest features anyway.
So these are my 5 cents. I keep an eye on the mailing list and will try to reply in timely manner to any questions.
Bartłomiej
[1] http://github.com/archlinuxarm _______________________________________________ arch-ports mailing list arch-ports@archlinux.org https://lists.archlinux.org/listinfo/arch-ports