IRC channel logs
2025-02-04.log
back to list of logs
<oriansj>shells are weird, in that some ways they shouldn't be used for programming but they feel close to what one might want (to the point that things like powershell/nushell make sense) but also are missing the unified design which would put their output into a format that would eliminate errors (like having proper database table outputs) but there have been a couple shell as spreadsheet projects out there which hint at that idea but fail to <oriansj>embrace the power that could come with it. <oriansj>probably because too much of the world refuses to comply to such a strict standard <oriansj>but perhaps I look at the 1+GB memory footprint of nushell and look at the 13.1MB footprint of havoc (a wayland st port) and the 16MB footprint of bash and think: jesus shells are bloated (even kaem is 80KB when built by M2-Planet) <lanodan>16MB for bash? The static build with glibc? <lanodan>At least mksh here sits at 361K static and that's not even size optimized and there's readline-like prompt included. <mid-kid><stikonas> maybe worth switching in your instructions too? <mid-kid>ninja isn't really a problem with the gentoo bootstrap, meson is a slight bit more complicated due to how it and python are set up in gentoo/ebuild infrastructure. <mid-kid>and neither is re2c - the gentoo bootstrap doesn't really go to great lengths like live-bootstrap and debian do to regenerate all files. <mid-kid>I've tried running samurai on my gentoo systems, it ended up segfaulting too often. <mid-kid>So I switched back. But that was 0.2, it's probably better now. <ztrawhcse>oriansj: I wonder how you got such a huge bash build? <aggi>maybe he was referring to total memory usage reported by kernel for the entire system <aggi>by coincidence, my linux24/tinycc system boots up with just that amount of memory used; for kernel, terminal multiplexer, a few shell prompts and vis editor <lanodan>Anyone here got riscv64 support for alloca.S in tcc? It's missing in mob and I'm not familiar with riscv assembly. <matrix_bridge><Andrius Štikonas> fossy: I think rootfs.py might be broken on older python, e.g. Ubuntu 22.04 <matrix_bridge><Andrius Štikonas> lanodan: no, I think we tried to avoid alloca before GCC in live bootstrap <matrix_bridge><Andrius Štikonas> fossy, line 58 with fstring upsets python 3.10 <lanodan>And seems like cproc has alloca so yeah, good enough for me <aggi>salvaged some ancient ghostscript 7.07 version and mupdf for framebuffer console, with tinycc/static-linking; at least it does pass compiletime and pdf can be rendered <aggi>it's just too much effort, ~500 builds plus kernel plus libc forked for tinycc support <aggi>to do all runtime testing and bugfixing <aggi>nonetheless, it's feasible, and worth the effort,... i'll have a look at some TeX implementation next <aggi>i've contacted some german bank, known as "eco bank" (umweltbank), maybe some contracting can be arranged with them <aggi>which would simplify things, if i could just git-push all stuff and not worry about any business model <aggi>currently, i am a little hesitant to publish the musl-libc fork for example (it's residing inside an uncommitted git-workspace still) <aggi>i'll know soon, if and how i could continue work, i've just received some more blackmail from bureaucracy <aggi>as far as tinycc/static-linking support was concerned fort a complete system profile including development utilities, that's a 100% mission accomplished already <aggi>all else is regular maintenance and cleanup-tasks now, to maintain a distribution built upon this