| 26 Apr 2026 |
 Lotte (it/its)/Cinny (she/her) θΔ& | cross compiled for running on riscv | 20:35:14 |
 alexfmpe | the correct Alex will now take over | 20:35:44 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i think i just recreated the hadrian patch in a worse way | 20:35:55 |
 Alex | Please do share how if you manage to get that to work. I couldn't convince Hadrian to produce the correct output. | 20:35:58 |
| Alex | Also, you may want to double check that the binary is actually for running on riscv64. Hadrian has a tendency to build a stage 2 cross-compiler instead of a stage 1 cross-compiler. | 20:37:04 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i also got a bootstrap from a debian 9.6.6 package | 20:37:41 |
| Lotte (it/its)/Cinny (she/her) θΔ& | it's still in stage 0 | 20:37:52 |
| Lotte (it/its)/Cinny (she/her) θΔ& | lemme look at the object files though | 20:38:19 |
| Alex | Ah, if you're booting a fully native build from Debian's GHC it might be fine. I haven't touched Debian's packages. | 20:38:42 |
| Alex | * Ah, if you're booting a fully native build from Debian's GHC it might be fine. I haven't tried Debian's packages. | 20:38:51 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i was doing it before but i replaced it with a bootstrap from cross compiled ghc9.4.8 | 20:39:32 |
| Lotte (it/its)/Cinny (she/her) θΔ& | which is what i originally wanted to do but was foiled by stdenv shenanigans | 20:40:04 |
| Lotte (it/its)/Cinny (she/her) θΔ& | 
Download asm file in build dir is riscv so it's probably fine | 20:41:07 |
| Alex | Nice. Looks promising. Since you mentioned 9.4.8 cross as a boot, is it a native build? | 20:42:24 |
| Lotte (it/its)/Cinny (she/her) θΔ& | the 967 build is native | 20:42:58 |
| Lotte (it/its)/Cinny (she/her) θΔ& | (running in qemu because my vf2 ain't got the cpu power for that) | 20:43:36 |
| Alex | Yeah, that should work without trouble. If it doesn't, I'd be curious to reproduce the issue and investigate why. | 20:43:46 |
| Alex | The VF2 can do it, provided you're willing to wait 20h for the GHC derivation alone. :D | 20:44:09 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i had an issue with 948 being single threaded which hadrian tried passing +RTS -qg to which single threaded 948 didn't understand | 20:45:48 |
| Lotte (it/its)/Cinny (she/her) θΔ& | fixed that though by patching out the flag (the incorrect thing to do seemingly) | 20:46:32 |
| Alex | True, I did have to patch that out of Hadrian but never actually took the time to upstream it into Nixpkgs.
It's a bit tricky but possible: https://github.com/AlexandreTunstall/nixos-riscv/blob/df26291108c656d9ec47d4d57caf31810c0498e3/modules/compilers/ghc.nix#L40 | 20:48:11 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i put it into postpatch for hadrian | 20:48:58 |
| Lotte (it/its)/Cinny (she/her) θΔ& | 
Download cursed | 21:09:00 |
 emily | just "the mangler"? in my day it was known as the Evil Mangler | 21:09:47 |
| emily | also, "a Perl script"? as the file extension clearly shows, it's a literate Perl script | 21:09:59 |
| Lotte (it/its)/Cinny (she/her) θΔ& | it's a ghc 5 era doc file | 21:12:59 |
| alexfmpe | it's literally a Perl script, what's the problem | 21:41:23 |
| emily | it's a literate Perl script, which is better because it's more academic | 21:41:38 |
| alexfmpe | not sure if woosh | 21:41:50 |
| alexfmpe | or getting one-upped | 21:41:55 |
