| 11 Feb 2025 |
 @dmjio:matrix.org | ❯ nix-shell -p closurecompiler --run 'closure-compiler -W QUIET < ./result/bin/simple.jsexe/all.js > out.js && du -hs ./result/bin/simple.jsexe/all.js out.js'
The compiler is waiting for input via stdin.
4.5M ./result/bin/simple.jsexe/all.js
3.7M out.js
| 00:07:26 |
| @dmjio:matrix.org | seems advanced opts aren't working with this version of closure compiler, even when externs are specified | 00:07:55 |
 alexfmpe | by "aren't working" do you mean incorrect behavior or just less savings that it used to? | 00:13:49 |
| alexfmpe | because I see use of ADVANCED in hello world example in https://gitlab.haskell.org/ghc/ghc/-/issues/24584 | 00:14:37 |
| alexfmpe | haven't yet tried out closure on actual js backend stuff | 00:14:56 |
| alexfmpe | though I think we're in the wrong room to continue this conversation | 00:15:17 |
| @dmjio:matrix.org | -O ADVANCED > out.js gives me a 0B file. The version of closure compiler miso uses is old. This one I just pulled from nixpkgs 25 | 00:15:27 |
| alexfmpe | shall we move this branch to #ghc-js-backend:matrix.org ? | 00:15:41 |
| @dmjio:matrix.org | sure | 00:16:09 |
 Divam | I agree with maralorn; for all projects I am aware that still use ghcjs from nixpkgs (via the old reflex-platform), there won't be a migration to newer nixpkgs. | 04:38:20 |
 eldritchcookie | i've spent the better part of today to get fourmolu to compile on ghc 9.12 but one of the fixes need fourmolu version 0.18.0.0 but the latest one available from nixpkgs is 0.16.2.0 what is the solution? | 23:04:48 |
| eldritchcookie | because i intend to upstream my fixes | 23:05:52 |
 maralorn | Is that on the haskell-updates branch? Seems unlikely that that wouldn't have a newer version. | 23:09:29 |
| 12 Feb 2025 |
| alexfmpe | fourmulu_0_18_0_0 attribute is around as is 0_17_00 | 02:23:58 |
| alexfmpe | * fourmulu_0_18_0_0 attribute is present in hackage-packages.nix as is the 0_17_0_0 one | 02:24:29 |
|  Mr Mayhem changed their display name from Mississippi_metalhead_mailman_maybe_magnolia_mayhem to ಊೌೋᭂᜲௌ᩿ۣۜᴭ¹ᶿ꥟. | 03:03:54 |
 Alex | In reply to @eldritchcookie:matrix.org
i've spent the better part of today to get fourmolu to compile on ghc 9.12 but one of the fixes need fourmolu version 0.18.0.0 but the latest one available from nixpkgs is 0.16.2.0 what is the solution? Nixpkgs always has both the version recommended by Stackage (from whichever LTS resolver is tracked by Nixpkgs) and the latest version on Hackage (as of when it was last pulled from). The fourmolu_0_18_0_0 package is the latter, while the Nixpkgs default is to use the former because that's more likely to build on the default GHC. | 04:06:44 |
| Mr Mayhem changed their display name from ಊೌೋᭂᜲௌ᩿ۣۜᴭ¹ᶿ꥟ to 𒀱. | 04:10:37 |
 sterni (he/him) | angerman: https://gitlab.haskell.org/ghc/ghc/-/merge_requests/13932 somewhat interesting for haskell.nix | 15:03:28 |
| Mr Mayhem changed their display name from 𒀱 to 𒀱Lord Mayhem of the Magnolias. | 19:45:12 |
| 14 Feb 2025 |
| sterni (he/him) | alexfmpe: https://hydra.nixos.org/build/289434054 | 01:13:37 |
| 12 Feb 2025 |
 angerman | sterni: nice! We are moving away from Hadrian at IOG. | 21:26:05 |
| 14 Feb 2025 |
| alexfmpe | huh looks like there was a common jailbreak in master that got removed at some point during current PR
added it back to 9.12 overrides
https://github.com/NixOS/nixpkgs/pull/381916 | 01:39:04 |
| 15 Feb 2025 |
| alexfmpe | https://github.com/NixOS/nixpkgs/pull/382159 gets haskell.packages.ghc810.miso building again
however, I just checked and haskell.packages.ghcjs810.miso was already broken in master soooo...moot point? | 00:21:13 |
| alexfmpe | seems to be due several packages present in LTS moved to js backend support without keeping ghcjs compatibility | 00:24:13 |
| alexfmpe | switching cpu from "js" to "javascript" and what not | 00:25:38 |
| alexfmpe | I wouldn't bet on someone being motivated to work on adding backward compat to these packages rather than working on moving to 9.12 js backend | 00:28:32 |
| alexfmpe | so maybe ghcjs is no longer a reason to keep 8.10 around | 00:28:48 |
| alexfmpe | that is, in the hypothetical case of someone using ghcjs with recent-ish nixpkgs | 00:31:12 |
| maralorn | Ah, thank you for figuring that out. | 01:46:59 |
