IRC channel logs
2026-01-18.log
back to list of logs
<aleksih>Thank you, I appreicate the detailed answers <aleksih>The nixpkgs approach starts from stage0-posix and (for now) takes configure scripts et al for granted, and native x86_64 does seem to be pretty functional with a few patches, as well as riscv64 <aleksih>But yeah I suppose for live-bootstrap, we wouldn't have Fiwix for anything other than i386 <aleksih>Regardless I would be interested to see if I can help in any way wrt porting to AArch64 or improving the current arch support. We have some bugfix patches in nixpkgs for mescc and tinycc-bootstrappable, related to x86_64 and riscv64. Though I'm sure many, if not all, could already be fixed in draft branches <mwette>Is repo for MES still on savannah or has it moved? <ekaitz>mwette: it is in savannah but janneke and I have copies in gitlab and codeberg (respectively) <stikonas>I should probably upload stage0-uefi to codeberg too, so that it's not just on my home server... <aggi>apropos codeberg, since recently they have walled access to their services with javascript which blocks navigation with links/lynx/w3m etc. <stikonas>you just can't run git forge without any kind of AI blocker <stikonas>those AI scraping bots would probably use something like 1 TB / month on my home internet connection <stikonas>hmm, just realized it will be annoying for us to (at some point in the future) migrate bootstrappable repo to sha2 git repos <aggi>few days ago i've git-tagged and signed repos required for the tinycc-driven distribution <aggi>considering upload/gitweb for this, i've checked the size of this: it's ~9GiB distfiles and git repositories in total :( <aggi>if i did another cleanup of this maybe i can cut this in half maybe <lanodan>Reminds me that so far I plan to keep on using sha1 for interoperability reasons (and git getting rusted strengthen that part a lot) <aggi>a i586-tcc-linux2-musl.iso release is ~1.8GiB in size (statically linked with debugging symbols, so this could be cut in half too without debugging symbos) <aggi>well, git-stability itself is a whole other story - OpenBSD are working on a re-implementation of it already with their got fork <aggi>i'm considering to shrink the package set again, for some minified dev-release, but i intentionally kept as much as possible initially <aggi>when sanitizing the dependency graph from c++ seeing to how much could be salvaged, and to test i386-tcc with as many builds as possible <aggi>next i'll have to cope with a few bureaucratic tasks anyway, and i didn't yet migrate my own main development host back onto linux-2 yet for further runtime-testing <aggi>thinking of git again, while ago, for fun i grabbed some ancient slackware11.iso which shipped with some older git-version <aggi>regardless of sha2-transition there's been compatibility breakage before (this old git-version could not cope with reposities created with new git-versions, both using sha1) <aggi>and, i've noticed a few other quirks already; upstream verious distfiles zipped their distfiles with some new compression (forgot which it was) <aggi>i couldn't yet support this new compression with a no-c++/tinycc driven system, it's just annoying, worst case i'll manually repack distfiles <aggi>various other builds transitioned from GNU/autotools (which wasn't trivial to support with tinycc already) to python/meson (i think iputils, i won't pull in meson) <aggi>some other builds transition towards cmake (that's c++, i don't want to nor can't pull that either) <aggi>overall, up until feb/2025 (that's the last time i synced against latest ~testing from gentoo-portage), the system profile for no-c++/tinycc is complete still <aggi>including full cross-compilation support and static linking