| 26 May 2021 |
 Gytis Ivaskevicius | We were talking about ion shell a little - got an email about this :D https://gitlab.redox-os.org/redox-os/ion/-/merge_requests/1157 | 11:32:40 |
 andi- | In reply to @toonn:matrix.org
andi-: You suggest seed->GCC8->GCC9->GCC10, right? What I'm saying is if the seed is capable of building GCC9 but not GCC10 you don't need to keep the entire chain. You'd do seed->GCC8, start again from the seed->GCC9->GCC10. Oh yes of course. As long as you can skip a version you should do it. | 11:34:58 |
 toonn | Oh ok, that wasn't obvious to me from what you were saying. | 11:35:24 |
| andi- | well it is common for compilers such as rustc to always require the previous version. | 11:38:47 |
| andi- | GCC might be a different story as they actually have a language spec.. | 11:39:01 |
| toonn | Require? That's unfortunate. Thank god GHC doesn't do that. | 11:46:55 |
| andi- | Can you still boot GHC from bash or C ? | 11:56:29 |
| toonn | That I do not know. Sounds like something that's more likely to have bitrotten than not. | 11:57:21 |
| andi- | I remember attending a Haskell hackathon where you had to provide a compiler and then you would bootstrap GHC in like 3 stages | 11:58:41 |
|  justinrestivo joined the room. | 12:27:43 |
| andi- | Have we actually bothered with reproducibility on MacOS before? So far every single python build is different... | 21:05:02 |
|  samueldr changed their display name from samueldr‎ to samueldr. | 21:05:27 |
 baloo | I think only linux is in the scope | 23:19:56 |
| baloo | r13y only tests for linux | 23:20:30 |
| 27 May 2021 |
 NinjaTrappeur | In reply to @andi:kack.it
Can you still boot GHC from bash or C ? You currently can't :( | 06:29:37 |
| NinjaTrappeur | https://www.joachim-breitner.de/blog/748-Thoughts_on_bootstrapping_GHC | 06:30:14 |
 siraben | x-post from Melg8: "If there somebody interested in nix bootstrap, please checkout https://github.com/NixOS/nixpkgs/issues/123095 i think we need some discussion/thoughts about it. " | 06:33:13 |
| NinjaTrappeur | I remember reading a WIP based on hugs but I can't find the link anymore :( TL;DR: there's been some effort in that area, but as far as I can tell, nothing beyond a WIP. | 06:33:22 |
| siraben | the only fully bootstrappable Haskell compiler I'm aware of right now is https://github.com/OriansJ/blynn-compiler which implements a subset of GHC's Haskell (enough for it to self-host and be accepted by GHC) | 06:34:38 |
| siraben | but GHC itself has lost a bootstrap long ago | 07:24:46 |
|  adisbladis joined the room. | 07:44:58 |
| 28 May 2021 |
|  adisbladis left the room. | 00:39:19 |
| 29 May 2021 |
|  Ochoa joined the room. | 02:35:52 |
 immae | https://r13y.com/ still mentions #nixos-r13y (cc grahamc (he/him)) | 09:46:34 |
| immae | (and congratulations for managing to remove gcc from the unreproductible list btw!) | 09:48:49 |
 @grahamc:nixos.org | Can you send me a pr? Grahamc/r13y.com I think | 11:29:11 |
| siraben | In reply to @immae:matrix.org
(and congratulations for managing to remove gcc from the unreproductible list btw!) yay! | 11:44:58 |
| siraben | what was the effect on performance of compiled binaries? | 11:45:16 |
| andi- | In reply to @siraben:matrix.org
what was the effect on performance of compiled binaries? Compilation is now about 10% slower IIRC | 11:46:07 |
| siraben | oh, and are the binaries slower? | 11:46:50 |
