IRC channel logs

<stikonas>it turns out its not easy to build other shells

<lanodan>Well oksh seems trivial to build (at least can just be built with a single tcc invocation, no specific buildsystem needed) but yeah the other shells are a pain

<aggi>lanodan: oksh too was much easier to backport onto linux-tcc/musl than bash-5.x

<aggi>i did not succeed however to swap bash with oksh for gentoo-portage

<aggi>a whole other story, since i used portage for dependency-tracking when hacking tcc-toolchain support

<lanodan>Yeah, after all the packaging spec depends on bash.

<aggi>yep, with live-bootstrap configurator, but i think this one is not plagued by bashism like gentoo-portage is

<aggi>so, from that perspective, swapping bash for any other shell such as oksh is much less troublesome

<aggi>and, just as said, back-porting oksh succeeded here... it's my main system shell

<aggi>on linux-2.4.37.11-tcc

<lanodan>By packaging spec I mean the gentoo one that portage, pkgcore, paludis, … implements

<aggi>yes, that's the major burden, and the reason why i want to re-write ALL gentoo-portage/ebuild packaging

<aggi>for the live-bootstrap configurator steps/ system

<aggi>for all ebuilds that passed compile/run-time with tcc-toolchain

<aggi>~500 total

<matrix_bridge><Andrius Štikonas> Well, perhaps oksh would be a good match for early bootstrap

<matrix_bridge><Andrius Štikonas> But of course somebody has to try and test

<lanodan>Yeah, plus like there's probably quite few bashisms here and there in upstream code, specially stuff required to regenerate files / bootstrap

<lanodan>Although interestingly only one I hit in my pretty minimal bootstrap-initrd is iproute2 depending on GNU Make, which so far I replaced with a muon-compatible meson.build (but muon's rather big and complex to review so thought about replacing the Makefiles instead)

IRC channel logs

2026-08-10.log