IRC channel logs
2026-03-09.log
back to list of logs
<aggi>seems to explain why neither pnut nor tcc_bootstrap_alt weren't merged into live-bootstrap master with some rootfs.py config switch for this <aggi>both have the advantage of faster compilation of tcc, but pnut was not yet fully refined for integration and needs mes/libc.a anyway <aggi>and live_bootstrap_alt depends on several different tcc-versions <stikonas>well, tcc_bootstrap_alt just have so many steps, it will be very hard to port it to other arches <stikonas>so far mes is far more cross-platform compared to others <aggi>admitted; tbh however, i find LISP/scheme difficult to read, but that's my problem <stikonas>yes, I find it a bit difficult to read too, but I still managed to make some changes to mescc <stikonas>it's just very unusual if you come from C background <aggi>i too have in mind still, one day maybe live-bootstrap will arrive on free/open hardware with an FPGA (ao486 and similar) <aggi>it's not specific to x486, but with such development hardware resource constraints are significant <aggi>+10minutes compile-time for MES/mes-cc/tcc is a significant bite taken on modern hardware <aggi>hence it's good bootstrap_alt and pnut exist as fallback option <LaurentH>aggi: live-bootstrap with pnut makes mes/libc.a from tcc-pnut (which is compiled with pnut's libc). The 2 different libc implementations make it a bit confusing but that's needed because pnut-exe doesn't support inline M1 assembly like Mes so it needs a libc that works with the built-in syscalls it has <stikonas>well, I think gtker is working on GAS compatible assembler in the background <stikonas>so at some point we might switch away from M1... <stikonas>well, it will still be needed for early assembly stages <mwette>stikonas: I released a 3.04.1 (tarball) to github and savannah <stikonas>back home now, though with a bit of delay... <mwette>lemme know if any more problems pop up <aggi>it's late, i'm certain both pnut and default mescc will work, but it has to wait until tomorrow; good night. <stikonas>mwette: I think it's fully working now with mes (with those few mes patches that were discussed yesterday) <stikonas>but that's with janneke now, I think he merged them and they'll be in the new release <stikonas>running gen-c*-files with mes is somewhat slower than I remember it in version 1.00.2 of nyacc but I guess that's expected <stikonas>and perhaps ekaitz's work will eventually speed up mes a bit <stikonas>hmm, actually building tcc with mescc failed, but that might be mes problem, or maybe I need to rebuild mes with mescc... <stikonas>anyway, that's investigation for another day <stikonas>but building meslibc with mes-m2/mescc/nyacc-3.04.1 worked <aggi>it's not been much done today, tcc-0.9.26.x86.checksums produced from pnut match