IRC channel logs

<oriansj><[exa]>: it should be possible to convert blynn to output M1 assembly and then you could do a haskell bootstrap avoiding C entirely.

<oriansj>stikonas: unfornately, no I haven't had a chance to look into yet. Thank you for the reminder.

<oriansj>ekaitz: congrats on the upcoming kid. (Just remember the work can wait, the kid definitely can't)

<ekaitz>oriansj: hopefully with his help i'll be able to make more work in less time later hehe

<ekaitz>thank you

<ekaitz>:)

<kpg>Hi

<kpg>Has anyone tried to bootstrap Kotlin the language?

<oriansj>ekaitz: it'll be about 8 years (from what I heard) before you can get them up to speed in software engineering to be able to help.

<ekaitz>:)

<ekaitz>kpg: i think some guix people did, julien i think

<ekaitz>kpg: i think the irc nickname is roptat

<oriansj>kpg: as I understand it you need to build openjdk-8-jdk and openjdk-11-jdk before you can bootstrap kotlin (and the path for that is in guix)

<kpg>ekaitz: thanks

<stikonas>openjdk is easy to build

<stikonas>or relatively easy

<oriansj>(especially since it is after GCC)

<stikonas>it's the further java stuff later that is hard, e.g. maven

<ekaitz>i think kotlin requires several versions to be build in chain

<kpg>I saw kotlin 0.5.1 required the jetbrains jdk... haven't tried if it will work with the common oracle/openjdk.

<oriansj>kpg: it is always possible for specific versions of a language to become "unbootstrappable"

<oriansj>then someone has a bunch of work to do to fix that. (Like we had with Bison)

<matrix_bridge><Andrius Štikonas> hmm, main guix tree doesn't seem to have kotlin

<stikonas>but maybe there was some work somewhere

<stikonas>I seem to recall somebody building some old version

<ekaitz>dude i'm pretty sure somebody worked on this

<stikonas>ok, found the repo https://github.com/ebourg/kotlin-bootstrapping

<ekaitz>aaah yeah there it is, roptat is there in the bottom

<kpg>great find, thank you stikonas

<stikonas>oriansj: as for bootstrap seeds I just spotted extra 0 in the comment, need to remove that

<stikonas>#:write ; (0x80480A08) should be #:write ; (0x80480A8) I think

<oriansj>stikonas: thank you for working to help shrink the root seeds.

<stikonas>yeah, let me push an update first

<stikonas>just a comment update

<stikonas>oriansj: ok pushed

<stikonas>oriansj: anyway, most of the work was done by goldstein before

<stikonas>I just ported it to x86_64

<oriansj>benefits of learning as a community ^_^

<stikonas>for x86, I just switched to a shorter mov instruction (1 byte)

<stikonas>+ 8 byte hack with the headers

<stikonas>for x86_64 I didn't overlap the headers though

<stikonas>it does work and run but probably against the spec

<stikonas>at least readelf complains

<stikonas>64-bit program header has a different order of the fields, so attempting to overlap them requires setting number of section headers to 1

<stikonas>we can keep e_shentsize to 0, so kernel ignores non zero e_shnum

<stikonas>but that seemed a bit too hacky

<stikonas>on the other hand, on x86 it seems to all work fine

<stikonas>just found this talk by rickmasters https://www.youtube.com/watch?v=ycLdhqn5VI4

<Mikaku>stikonas: good find and thanks for share it!

<[exa]>oriansj: true, thanks for the idea

IRC channel logs

2024-11-09.log