8 Oct 2024 |
alexfmpe | And saying goodbye to ols ghcjs | 15:01:55 |
alexfmpe | * And saying goodbye to ol' ghcjs | 15:02:07 |
MangoIV | Redacted or Malformed Event | 15:02:30 |
MangoIV | In reply to @alexfmpe:matrix.org And saying goodbye to ol' ghcjs IOG Stil funding this? | 15:03:11 |
alexfmpe | Think they stopped funding right before the finish line, but someone picked up the torch | 15:03:48 |
alexfmpe | https://blog.haskell.org/report-of-js-code-minification/ | 15:04:15 |
alexfmpe | 9.12 is looking pretty plausible for the switch | 15:04:50 |
maralorn | In reply to @mangoiv.:matrix.org IOG Stil funding this? I mean it’s the guys from IOG who were writing it. And they basically still do but now on a much reduced time budget. | 15:06:59 |
| schuelermine changed their profile picture. | 16:32:15 |
9 Oct 2024 |
doyougnu | In reply to @mangoiv.:matrix.org IOG Stil funding this? The JS backend or GHCJS? We have stopped working on GHCJS altogether. | 10:53:03 |
doyougnu | Regarding the Linker, yea we know its slow and memory hungry. There is a ticket for this https://gitlab.haskell.org/ghc/ghc/-/issues/25324, one of the major reasons is that the linker is actually where the JS concrete syntax is being generated. But in all honesty we haven't spent serious cycles trying to improve it yet because we had different priorities come down the pike. | 10:55:57 |
doyougnu | In reply to @alexfmpe:matrix.org Think they stopped funding right before the finish line, but someone picked up the torch Yea this is Serge S. Gulin he's been a hero in the space | 10:56:43 |
MangoIV | In reply to @doyougnu:matrix.org The JS backend or GHCJS? We have stopped working on GHCJS altogether. Yea I thought you dropped it in favour of the new backend so that matches my expectations. | 11:04:28 |
maralorn | In reply to @doyougnu:matrix.org The JS backend or GHCJS? We have stopped working on GHCJS altogether. Ah, I didn’t map you to IOG.^^ | 11:18:27 |
maralorn | In reply to @alexfmpe:matrix.org https://blog.haskell.org/report-of-js-code-minification/ I got a very helpful clarification E-Mail from Luite on the situation regarding code size, I embedded that information into my WIP document here: https://md.darmstadt.ccc.de/x3lqwjf4QXaTdBURWoWfhQ?view | 13:40:20 |
maralorn | If anyone has more details about the state of JS and WASM backend in nixpkgs which I can put into that doc, please let me know. | 14:17:44 |
emily | In reply to @maralorn:maralorn.de I got a very helpful clarification E-Mail from Luite on the situation regarding code size, I embedded that information into my WIP document here: https://md.darmstadt.ccc.de/x3lqwjf4QXaTdBURWoWfhQ?view "plattform" seems like a typo | 14:51:01 |
emily | nice doc btw :) | 14:51:07 |
maralorn | Huh, you know. I come from the North of Germany where some of the old people still speak a language called "platt". | 14:54:46 |
| schuelermine changed their profile picture. | 23:46:57 |
10 Oct 2024 |
Serge S. Gulin | Right now I am on https://gitlab.haskell.org/ghc/ghc/-/merge_requests/13185 still, it took more time than I expected. It raised many speculative questions from my side but I am in progress with it and something is looking clear.
These questions are not about that one optimization. It is more broad: role of sinker, should we extend linker with move advanced dead code elimination, should we extend ghc optimizer with some things from sinker.
Take a look at the discussion at the link. Howbeit I am in progress with it and hope soon something will be merged and some decisions will be discussed. | 07:56:58 |
Serge S. Gulin | Regarding reflex, well, it compiles https://github.com/reflex-frp/reflex-dom/pull/474#issuecomment-2310097582 | 07:59:16 |
Serge S. Gulin | https://github.com/odr/reflex-js Typeable said they got it compiling there, I was unable to reproduce it at Darwin ARM, but at Linux it should be fine. Need take deeper look what went wrong there. | 08:00:27 |
Serge S. Gulin | * https://github.com/odr/reflex-js Typeable said they got it compiling there, I was unable to reproduce it at Darwin ARM, but at Linux it should be fine. Need take deeper look what went wrong there. | 08:00:34 |
Serge S. Gulin | I planned to make reflex work but stuck with https://gitlab.haskell.org/ghc/ghc/-/merge_requests/13185 | 08:01:10 |
maralorn | Serge S. Gulin: ❤️ | 08:07:17 |
maralorn | I am mostly curious about the stuff in GHC because everything in the libs is probably something I would know how to fix. | 08:09:16 |
11 Oct 2024 |
emily | would appreciate a Haskell maintainer taking a glance at https://github.com/NixOS/nixpkgs/pull/346928 | 00:19:15 |
emily | it targets staging because it depends on a PR that was just merged | 00:19:32 |
emily | so it can't go to haskell-updates | 00:19:36 |