| 9 May 2025 |
 lillecarl | I just posted a forum thread regarding flake-compat which some might find interesting: https://discourse.nixos.org/t/flake-compat-as-alternative-to-nix-flakes/64029 It discusses using flake-compat instead of npins essentially, and while you still have to do "flake gymnastics" to output for all systems it evaluates from the filesystem instead of the store which is pretty cool and might be useful to some while fully flake compatible. I'm off to bed but please have a read! Might be something someone's already doing but I thought it pretty dang cool. I'm off to bed now however 😄 | 01:34:15 |
| lillecarl | * I just posted a forum thread regarding flake-compat which some might find interesting: https://discourse.nixos.org/t/flake-compat-as-alternative-to-nix-flakes/64029 It discusses using flake-compat instead of npins essentially, and while you still have to do "flake gymnastics" to output for all systems it evaluates from the filesystem instead of the store which is pretty cool and might be useful to some while fully flake compatible. I'm off to bed but please have a read! Might be something someone's already doing but I thought it pretty dang cool. 😄 | 01:56:10 |
| lillecarl | * I just posted a forum thread regarding flake-compat which some might find interesting: https://discourse.nixos.org/t/flake-compat-as-alternative-to-nix-flakes/64029 It discusses using flake-compat instead of npins essentially, and while you still have to do "flake gymnastics" to output for all systems it evaluates from the filesystem instead of the store which is pretty cool and might be useful to some while fully flake compatible. Might be something someone's already doing but I thought it pretty dang cool. 😄 | 15:46:36 |
| lillecarl | Lix had already implemented the feature I patched on edolstras flake-compat. I extended theirs further so it also doesn't copy inputs to store if they're of the "path" type. https://git.lix.systems/lix-project/flake-compat/pulls/79 😄 | 15:53:24 |
| 10 May 2025 |
|  Viktor Vilhelm Sonesten joined the room. | 16:39:38 |
| Viktor Vilhelm Sonesten changed their display name from Viktor Sonesten to Viktor Vilhelm Sonesten. | 16:41:28 |
| 13 May 2025 |
|  kraem changed their profile picture. | 13:51:39 |
| 14 May 2025 |
 markuskowa | How about Wednesday 29th of May for the next meeting? Thumbs up. Comments. | 06:34:53 |
| kraem changed their profile picture. | 17:00:45 |
| kraem changed their profile picture. | 17:01:54 |
| kraem changed their profile picture. | 17:07:10 |
| kraem removed their profile picture. | 17:10:49 |
| kraem set a profile picture. | 17:17:59 |
| 16 May 2025 |
|  Edward Tjörnhammar joined the room. | 16:40:52 |
| 20 May 2025 |
| markuskowa | Since there was no feedback for next Wednesday, let me suggest Wednesday, June 4th instead for the next meet-up. | 07:27:43 |
| 22 May 2025 |
|  @309avklarat:matrix.org changed their display name from 309 to Serakhtis. | 08:21:33 |
|  bitops joined the room. | 10:48:17 |
| 25 May 2025 |
| markuskowa | It looks like we have critical mass. Let's meet on Wednesday June 4th, 18:00 at the Nordic Light hotel (Vasagatan 11, Stockholm). | 09:56:18 |
|  @fgaz:matrix.org left the room. | 18:13:41 |
| 29 May 2025 |
|  @deeok:matrix.org joined the room. | 16:58:47 |
| 1 Jun 2025 |
| lillecarl | I must say, vibecoding is 10/10. I've got a derivation creating a Python library from the grpc proto for the Kubernetes CSI spec and an empty implementation of the CSI interface in Python. Essentially by iterating with GH Copilot 😄 | 03:35:57 |
| 2 Jun 2025 |
| markuskowa | Quick reminder: we meet on Wednesday, 18:00 at the Nordic Light hotel. | 11:38:31 |
 I use NixOS btw | In reply to @lillecarl:matrix.org
I must say, vibecoding is 10/10. I've got a derivation creating a Python library from the grpc proto for the Kubernetes CSI spec and an empty implementation of the CSI interface in Python. Essentially by iterating with GH Copilot 😄 Yep, it's getting better. I just wish LLMs would say my ideas are stupid when they are stupid :/. It's too easy to be led astray.
I tried Gemini a lot last month (trial) and it's impressive but it still needs to be babysitted,
The age old principle to only copy code I fully understand is still relevant imo | 11:50:24 |
 dish1892 | In reply to @ahbk:matrix.org
Yep, it's getting better. I just wish LLMs would say my ideas are stupid when they are stupid :/. It's too easy to be led astray.
I tried Gemini a lot last month (trial) and it's impressive but it still needs to be babysitted,
The age old principle to only copy code I fully understand is still relevant imo Have you tried mistral? Imo it’s better in telling you when it doesn’t know. I’m using Continue with the free tier of mistrals api and are pretty happy | 12:00:40 |
| I use NixOS btw | I'm fine with them guessing though, keeps my brain from turning into mush 😅
But they're not critically evaluating my requests, and worse, just blindly affirms them. This has led to numerous misguided endeavors for me lately | 12:24:26 |
| lillecarl | In reply to @ahbk:matrix.org
Yep, it's getting better. I just wish LLMs would say my ideas are stupid when they are stupid :/. It's too easy to be led astray.
I tried Gemini a lot last month (trial) and it's impressive but it still needs to be babysitted,
The age old principle to only copy code I fully understand is still relevant imo Agreed, but it's a lot easier to evaluate if something is right than coming up with it yourself :) | 19:50:31 |
| lillecarl | I have acess to many LLMs through Kagi but the copilot chat has been good for the price, it won't discuss politics or write poems though 🤪 | 19:54:13 |
| 4 Jun 2025 |
| I use NixOS btw | Absolutely, and it feels like an efficient way to learn relevant stuff around the problem as well | 06:57:51 |
| I use NixOS btw | In reply to @markuskowa:matrix.org
Quick reminder: we meet on Wednesday, 18:00 at the Nordic Light hotel. I can't join this time due to my poor planning :( next time! | 06:59:12 |
| bitops | Planned to join you but I've been delayed, will anyone still be there around 19:00? | 16:15:18 |
