IRC channel logs

2024-08-25.log

back to list of logs

<nimaje>as I just saw go-php, a go interpreter written in php, maybe that can be used to bootstrap go, is php already bootstrapped? https://github.com/tuqqu/go-php
<AwesomeAdam54321>Is there a reason Go can't be bootstrapped from the last version written in C?
<nimaje>well, more bootstrap paths are good (and cosinusoidally found something that looks like a pregenerated file, no idea how hard it would be to regenerate)
<matrix_bridge><Andrius Štikonas> Well, that's just bison file
<matrix_bridge><Andrius Štikonas> So no big deal
<matrix_bridge><Andrius Štikonas> And you can go from 1.4 to maybe 1.17 and then modern go
<matrix_bridge><Andrius Štikonas> Or gccgo as discussed above...
<matrix_bridge><Andrius Štikonas> @irc_libera_andydude:stikonas.eu: v10.0.1.tar.gz is probably what live bootstrap downloads
<matrix_bridge><Andrius Štikonas> It could be some git archive
<oriansj>andydude: it might not be a package name but it is what the upstream uses: steps/kexec-tools-2.0.22/sources:https://github.com/horms/kexec-tools/archive/refs/tags/v2.0.22.tar.gz af618de7848142f204b57811f703de3ae7aa3f5bc5d52226db35800fa8fc4dff
<oriansj>perhaps if you make a patch to live-bootstrap ensuring more clear package names; it would eliminate such poorly named tarballs from the source list? and then there would be no need to have such names in the sources repo nor the mirror server.
<matrix_bridge><cosinusoidally> @irc_libera_awesomeadam54321:stikonas.eu: Go 1.4 is written in C and can be used to bootstrap
<lanodan>Is there a reason to stick to those names though? All distros I've seen would rename it to kexec-tools-2.0.22.tar.gz as that would also match the top-level folder.
<oriansj>lanodan: probably because no one got to it yet.
<oriansj>the usually is the reason for most problems in community projects; no one was annoyed enough to say "fuck that" and fix it and keep it fixed.
<lanodan>yeah, been there myself many times
<oriansj>and usually once it is done just enough to get you to good enough, either you lowered the effort for someone else to carry it forward or it just stops; hence why having as big of a community as possible pushing in the same general direction is essential to stop things from getting stuck.
<oriansj>like right now, unxz in mescc-tools-extra is broken for RISC-V (64bit) but I just can't work up the motivation to dig into the exact reason why its breaking but it is working for x86 which I do use.
<oriansj>(I am guessing having to do with the top bit behavior in M2-Planet for RISC-V constants)
<matrix_bridge><Andrius Štikonas> Also you can't rename upstream tarballs
<matrix_bridge><Andrius Štikonas> So renaming just in mirror makes it more complicated
<matrix_bridge><Andrius Štikonas> And unxz is only broke for m2-planet, other compilers compile it fine
<matrix_bridge><Andrius Štikonas> So yes, there is less motivation to fix it
<matrix_bridge><Andrius Štikonas> As it wouldn't unblock much...