IRC channel logs

<aggi>spotted yet, another, third regression in tcc

<aggi>which i think, it's almost impossible this couldn't have passed unnoticed

<aggi>situation is this, while #bootstrappable may prefer to stay at a stable release tagged for tcc

<aggi>i rather keep the option to sync with latest git HEAD of it, while regression-testing tcc against the i486-tcc-linux-musl.iso system profile that i got

<aggi>but i am loosing days and weeks and month bisecting through different versions and branches for several hundred builds i intend to keep on maintenance for this

<aggi>as of now, some regression got introduced into assembly parser of tcc that causes it to error out inside #comment sections within .S assembly files

<aggi>and another one that potentiall gets tcc stuck inside an endless loop while parsing options

<aggi>no big deal, but i am a little worried tcc branch becomes destabilized

<aggi>and I am not sure if all this belongs onto tinycc-devel mailing list nor into #bootstrappable channel

<lanodan>Well might make sense to push for something like a stable tcc branch and have that be the HEAD given how mob works

<aggi>the last release tag made for tcc is 7 years old, and i fear since then a few instabilities got introduced, which weren't spotted by regression testing

<aggi>that is since then, who knows, almost 200 commits to diff/bisect, if anytime 0.9.28 got tagged, and all regressions hit at once

<aggi>good news is, it seems possible still, on latest git head of tcc mob branch, to compile ~500 ebuilds with it, and linux-2.4.37.11 kernel

<aggi>i'll have to do a notable amount of cleanup tasks, then backporting stuff to linux-2.4 that i need

<aggi>because, i want to work and develop with such a pure tcc-linux system myself all day long, that's necessary for testing

<aggi>ideally, the "kernel bootstrapping" can be addressed, hence i'll keep tccboot in the test-matrix, although i wouldn't need a JiT/tccboot kernel

<matrix_bridge><cosinusoidally> Interesting, Slackware 12 (2.6 kernel) massively outperforms Slackware 15 (5.15 kernel) when running live-bootstrap in chroot mode. From 0 to gawk takes less than 2 mins when running my tcc_bootstrap_alt mode on the 2.6 kernel. It takes around 9 mins on the 5.15 kernel. The mescc mode takes around 21 mins on a 2.6 kernel. This is on an old i5.

<matrix_bridge><cosinusoidally> I still need to look at updating my live-bootstrap PR for mes-0.27, but I'm not sure when I will get around to that. I'm currently playing around with building things in tcc under fiwix.

<matrix_bridge><cosinusoidally> That is slackware under qemu

<matrix_bridge><cosinusoidally> bare metal mode under qemu takes around 20 mins for tcc_bootstrap_alt mode and 46 mins for mescc mode. I also need to look at optimising tcc_bootstrap_alt under bare metal mode.

IRC channel logs

2024-11-16.log