<fossy>mihi: i did successfully build coreutils for mes c library previously, however a few tools had to be omitted <fossy>and a large number of patches were needed <mihi>fossy, ok thanks. So I assume for busybox/toybox you would just need an extra number of patches :) <mihi>s/extra number/extra large number/ <fossy>mihi: i am guessing it would be a very complex job and it would be mostly disfunctional <mihi>I was toying with it yesterday and my first approach was to stub out all unsupported functions in libbb/platform.c. But when I reached stuff like rename and ctime (used by ls), I would agree that it would be mostly disfunctional. <mihi>I also tried some things with normal glibc (e.g. stubbing poll to always return 0) and at least for the shells it works. :) <mihi>not sure if bash also wants poll. <stikonas>fossy, mihi: do we actually need coreutils this early? I thought we can build gcc 2.95, glibc and then coreutils <stikonas>anyway, for now we have a few clear candidates that we need to build... <stikonas>patch, make, bash and try to get autotools <fossy>stikonas: we need coreutils for configure scripts.... <stikonas>well, for manual makefiles we can avoid those, but for autogenerated ones I guess you are right <stikonas>anyway, let's get patch/make/bash first :) <OriansJ>mihi: I will add any required primitives to M2libc with the hope that it solves the MesCC and M2-Planet libc problem. <OriansJ>It is just an incremental process; where something is discovered as being needed and then we add it. <OriansJ>yt: just to warn you I am going to do a force push to fixup the parallel make in a little bit. Sorry for the inconvience. <siraben>OriansJ: thanks for the merge, looks like blynn-compiler builds properly again <OriansJ>good we probably need to do some tweaks in the Haskell code when I pull out --bootstrap-mode to get you better performance but yt and I need to discuss #if defined to ensure M2-Planet work arounds are isolated and have a clean future elimination path. <OriansJ>I'm thinking variables like __M2_v1.x or _M2_Pointer_Arith but I'm sure yt has a better idea ***mephista is now known as spicy_icecream
***mephista is now known as spicy_icecream
<pder>siraben: i think i saw that too, however I think it is for crossly, and wont work with precisely. I think copying a good portion of precisely.hs up to that same point might be a good start for a prelude. <pder>Do you know of a good way to automatically include the prelude? <siraben>pder: Hm, looks like we'd need to expose some basic file primitives to Haskell so that we can read the prelude, then adjust the parser to add an includePrelude directive <pder>Do you know what the bottom part of precisely.hs is for? Starting with the line data CPP = CPPPass... <siraben>if you look at the cpp function it actually does the IO <fossy>gforce_de1977: yeah I realised how I broke it <fossy>I have a fix queued, ill push ur PR after I push bash <gforce_de1977>dumb question: kaem - i dont get the joke: how is the word formed/why? km? <bauen1>if you need a more complete libc before you can build gcc / glibc, you could try building musl with tinycc <bauen1>of course complex numbers won't work and you will need to patch around a few assembly instructions, but it is possible <bauen1>stikonas[m]: i'm using musl build using tinycc for myunix :D you do need to patch a number of big components out (like complex numbers) and some minor patches <gforce_de1977>bauen1: sounds good, can you point us to a patchset or something alike? <bauen1>gforce_de1977: for myunix v3 currently only exit() works :D i'll take a look at my earlier attempts that went further <gforce_de1977>bauen1: are you aware of, that 'bauen' means 'to build' in german? 8-) <bauen1>gforce_de1977: yes, the story behind my name actually goes back to my very first user name "on the internet" when i bought minecraft (which eventually lead to cpu building, then assembly and java) <bauen1>stikonas[m]: i'm pretty sure there's bugs in there too, best to start over <bauen1>stikonas[m]: in any case it boils down to: remove complex numbers (and maybe some other math related things), then rewrite some of the assembly that uses opcodes unsupported by tinycc (or implement them) <bauen1>stikonas[m]: keep in mind that myunix2 loaded raw binaries not elf files, leading to some pretty terrible things <fossy>keeps dying making temp files <fossy>luckily i've found an alternative solution for now <fossy>but we will need patch for linux kernel <stikonas>fossy: you mentioned that you will fix misplaced asterisk. or should I make a new PR? <fossy>stikonas: i'm almost 100% it's a missing macro, because guix compiles this successfully <fossy>make builds now, but needs testing <stikonas>if it doesn't work, it can be fixed in a followon patch <stikonas>hmm, patch tarball has very strange permissions too... <gforce_de1977>fossy: question: "catm config.h" ist just the same like 'touch config.h' ? <stikonas>gforce_de1977: yes, this is just my hack since we don't have touch <stikonas>actually, you don't need touch on normal system either <fossy>stikonas: i don't really like pushing things im not sure work to master, but i'll push it to a branch once this build finishes <stikonas>oh, maybe because I'm missing something in dev environment <stikonas>(but the error was error: macro 'geteuid' used with too many args) <fossy>no it's not, it's just something i added in dev environm,ent but not to the kaem script <fossy>and..... boom, pushed to patch branch typo fix <stikonas>once you said typo, I spotted it too and already tested <stikonas>fossy: also whitespace, please remove it <stikonas>version.c:27: error: 'PACKAGE_VERSION' undeclared <fossy>gosh, i must be very tired tonight <fossy>i did fix version.c too, but i didn't commit that change either <fossy>fwiw, PATCH_VERSION is not required <stikonas>with those changes it seems to at least run <fossy>yeah --version works, but when you try to use it to apply a patch it fails (at least inside live env) <stikonas>yeah ./patch: **** Can't create file /tmp/poizubgaerror 02: <stikonas>(my evening, probably very early morning for you) <fossy>stikonas: yeah thats what I think too <pder>fossy, is patch possibly failing because it it attempting to write to /tmp within the chroot and that directory doesnt exist? <stikonas>I had something similar in some of tar/gz/sed/diffutils <stikonas>I had to use it to to get creat function working <stikonas>pder: in any case patch should be able to create folders <pder>stikonas: that's great, thanks <stikonas>pder: the fix is on github review suggestions <stikonas>pder: and I've got bzip2 (conditional on pder work, so untested in live environment but it should work) <stikonas>and it's newer than bzip2 that I have on my current Gentoo system <stikonas>so I think we need to do patch -> make -> bzip2 and then we can do bash <stikonas>or maybe patch -> rebuild tcc with static patch -> make -> bzip2 -> bash <bauen1>stikonas: that's what happens when you have the us keyboard layout selected on a macbook air and try to switch to window 11 in irssi <bauen1>stikonas: the second time is when you can't believe that you're at fault <stikonas>fossy: yeah, I've seen something similar before <bauen1>xentrac: well, on this keyboard with the german layout when i want to type "/11", i use shift-7 to get the / which is & with the american layout apparently <bauen1>xentrac: alt-number only works for 0-9 :/ <stikonas>bzip2 already ships with hand-written makefile and only a small patch is needed to build with tcc <stikonas>and I tested my patch binary, it managed to apply necessary patch <fossy>stikonas: you made bzip using make or kaem? <stikonas>but I think the one you are bootstrapping will work <stikonas>that's why adding bzip2 will be just a few-liner <stikonas>fossy: but maybe once we have patch we can rebuild tcc first? <stikonas>or maybe after patch+make.... not sure what is easier <fossy>stikonas: right after patch, i have it queued up <fossy>stikonas: with that suggested change, it works fine, right? <stikonas>patch? yes, I tried to apply my multi-file bzip2 patch <stikonas>(and one of those files is actually created) <stikonas>so you can switch back to patching make too <stikonas>so, putting up WIP patches up for review works quite well <fossy>if you want to push your bzip2 to a branch and make a wip pr go ahead <stikonas>well, I can wait a bit, otherwise it will conflict with your changes to after.kaem.run <stikonas>then we can also start downloading tar.bz2 <xentrac>bauen1: on my irssi, alt-qwertyuio are the next 9 windows <bauen1>xentrac: thanks, that's good to know <xentrac>or is it some weird thing with my setup? <stikonas>like I said it's untested (can't test until make is there) <stikonas>(and I suspect that it needs rebuilt tcc because I didn't specify "tcc -static". Or alternatively I need to patch CFLAGS further...