| 2026-01-08 00:01:10 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 246 seconds) |
| 2026-01-08 00:03:43 +0100 | trickard_ | trickard |
| 2026-01-08 00:04:40 +0100 | <newmind> | any interest in a claude-code clone written in haskell, i'm currently working on something that's already supporting most of claude-code's features (minus a lot of polish) and a few extra things (llm tool generation) and would love some external feedback/ideas/contributions while it's still fairly early in development |
| 2026-01-08 00:07:40 +0100 | <ncf> | 1. no 2. fuck off |
| 2026-01-08 00:08:04 +0100 | <newmind> | alright? sorry? mind telling me what i did wrong here? |
| 2026-01-08 00:12:05 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 00:13:01 +0100 | timide | (~timide@user/timide) (Remote host closed the connection) |
| 2026-01-08 00:16:44 +0100 | <EvanR> | wow |
| 2026-01-08 00:17:02 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 265 seconds) |
| 2026-01-08 00:21:30 +0100 | newmind | (~newmind@91-133-90-252.dyn.cablelink.at) (Quit: Client closed) |
| 2026-01-08 00:21:57 +0100 | newmind | (~newmind@91-133-90-252.dyn.cablelink.at) |
| 2026-01-08 00:23:29 +0100 | Sgeo | (~Sgeo@user/sgeo) Sgeo |
| 2026-01-08 00:25:11 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) (Quit: xff0x) |
| 2026-01-08 00:25:29 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) |
| 2026-01-08 00:27:51 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 00:29:52 +0100 | marlino | (~marlino@96-8-193-27.block0.gvtc.com) (WeeChat 4.8.1) |
| 2026-01-08 00:29:52 +0100 | <haskellbridge> | <sm> llms and ai are extremely disruptive and people have strong feelings for and against |
| 2026-01-08 00:31:48 +0100 | <newmind> | certainly, and there are many ethical and societal considerations that come with that, for sure. doesn't mean they will be going away any time soon, nor that any use of them is automatically problematic though, right? |
| 2026-01-08 00:33:01 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 264 seconds) |
| 2026-01-08 00:35:24 +0100 | <mauke> | doubtful |
| 2026-01-08 00:37:21 +0100 | <jreicher> | newmind: it might be a bit like asking whether writing gambling software is automatically problematic. In theory, no, but... |
| 2026-01-08 00:40:07 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) (Quit: xff0x) |
| 2026-01-08 00:40:26 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) |
| 2026-01-08 00:40:30 +0100 | <newmind> | i think it would be closer to writing a game framework that can be used to create gambling software, but i get what you're saying from a certain point. especially in the commercialization of AI there's a lot going on that's "not good" to put it very mildly |
| 2026-01-08 00:43:18 +0100 | <jreicher> | Well when you say a clone of "claude-code" you're already specifying the use (code). And IMO LLMs/AI are deeply (and automatically) problematic for that. I quite like them for summarising a variety of prose sources though; I think that works quite well. |
| 2026-01-08 00:43:38 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 00:45:07 +0100 | <newmind> | well, i't an attempt to reign in the most problematic parts (arbitrary command execution, uncontrolled side effects) to a certain extend by running it through polysemy/effects. it's not automatically solving all the problems while still being useful, but it gives you at least some more control over whats running, and how |
| 2026-01-08 00:45:19 +0100 | Tuplanolla | (~Tuplanoll@88-114-88-95.elisa-laajakaista.fi) (Ping timeout: 240 seconds) |
| 2026-01-08 00:46:04 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) (Remote host closed the connection) |
| 2026-01-08 00:46:35 +0100 | L29Ah | (~L29Ah@wikipedia/L29Ah) (Ping timeout: 245 seconds) |
| 2026-01-08 00:46:41 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) DragonMaus |
| 2026-01-08 00:48:07 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 240 seconds) |
| 2026-01-08 00:48:34 +0100 | <jreicher> | Actually I should probably be more specific where code is concerned too. I strongly dislike using an agent to generate code. But using an agent for something like snippet completion is quite good; a bit like a more contextual language server. |
| 2026-01-08 00:48:58 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) (Remote host closed the connection) |
| 2026-01-08 00:49:35 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) DragonMaus |
| 2026-01-08 00:50:08 +0100 | newmind | (~newmind@91-133-90-252.dyn.cablelink.at) (Quit: Client closed) |
| 2026-01-08 00:50:19 +0100 | newmind | (~newmind@91-133-90-252.dyn.cablelink.at) |
| 2026-01-08 00:52:21 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) (Remote host closed the connection) |
| 2026-01-08 00:52:28 +0100 | <newmind> | personally, i'm more accepting when it comes to actual generation: i'm fine with an LLM generating functions and even modules, but not in a unsupervised manner. the code quality itself is too low, unstructured and has a tendency to take shortcuts instead of solving problems "the right way". if you constrain the types of a function a bit though, the |
| 2026-01-08 00:52:29 +0100 | <newmind> | results are actually quite usable |
| 2026-01-08 00:52:59 +0100 | DragonMaus | (~DragonMau@user/dragonmaus) DragonMaus |
| 2026-01-08 00:55:07 +0100 | <newmind> | but what i'm working on isn't really limited to coding per se. it's a TUI chat interface with an LLM backend and a set of tools that can be called (and run through polysemy, so just by the type signature you see what can even be accessed). |
| 2026-01-08 00:57:02 +0100 | Googulator | (~Googulato@2a01-036d-0106-4994-68db-cf64-05de-a70a.pool6.digikabel.hu) (Quit: Client closed) |
| 2026-01-08 00:57:19 +0100 | Googulator | (~Googulato@2a01-036d-0106-4994-68db-cf64-05de-a70a.pool6.digikabel.hu) |
| 2026-01-08 00:57:19 +0100 | <TMA> | it is a hit and miss in my experience. sometimes the result is usable or almost so, sometimes it "reasons" that it will ignore the instructions. "Do this." "Doing this is hard, let's do a simplified version instead." |
| 2026-01-08 00:57:28 +0100 | marinelli | (~weechat@gateway/tor-sasl/marinelli) (Remote host closed the connection) |
| 2026-01-08 00:57:29 +0100 | ChaiTRex | (~ChaiTRex@user/chaitrex) (Remote host closed the connection) |
| 2026-01-08 00:57:49 +0100 | marinelli | (~weechat@gateway/tor-sasl/marinelli) marinelli |
| 2026-01-08 00:57:53 +0100 | ChaiTRex | (~ChaiTRex@user/chaitrex) ChaiTRex |
| 2026-01-08 00:57:53 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) (Quit: xff0x) |
| 2026-01-08 00:58:11 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) |
| 2026-01-08 00:59:18 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 00:59:34 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) (Client Quit) |
| 2026-01-08 00:59:46 +0100 | <newmind> | yeah, and then it sticks to doing something in a convoluted way instead of taking an obviously simpler, more elegant and more universal solution... in this aspect it mimics a junior dev quite realistically, just without out the "able-to-learn" parts |
| 2026-01-08 00:59:47 +0100 | xff0x | (~xff0x@2405:6580:b080:900:4b0b:90a:cd82:2bd2) |
| 2026-01-08 01:03:55 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 240 seconds) |
| 2026-01-08 01:06:33 +0100 | ljdarj | (~Thunderbi@user/ljdarj) (Ping timeout: 252 seconds) |
| 2026-01-08 01:11:45 +0100 | <monochrom> | As an individual citizen of #haskell I say that I'm not interested in a Claude program but that's just me. As a moderator I say that "fuck off" is uncalled for. |
| 2026-01-08 01:13:46 +0100 | <geekosaur> | +1 |
| 2026-01-08 01:14:15 +0100 | <newmind> | it's perfectly fine, i fully get that it might be a charged topic, i was just wondering if i might have expressed myself in a way that's generally offensive >.> |
| 2026-01-08 01:15:05 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 01:15:14 +0100 | <monochrom> | I think you're doing fine. You just asked "anyone interested?" so it's fair game. |
| 2026-01-08 01:19:47 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 244 seconds) |
| 2026-01-08 01:25:43 +0100 | Googulator14 | (~Googulato@2a01-036d-0106-4994-68db-cf64-05de-a70a.pool6.digikabel.hu) |
| 2026-01-08 01:25:43 +0100 | Googulator | (~Googulato@2a01-036d-0106-4994-68db-cf64-05de-a70a.pool6.digikabel.hu) (Quit: Client closed) |
| 2026-01-08 01:27:49 +0100 | <newmind> | besides, the assessment seemed to be, at the very least, factually accurate :) |
| 2026-01-08 01:30:53 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 01:31:26 +0100 | trickard | (~trickard@cpe-50-98-47-163.wireline.com.au) (Read error: Connection reset by peer) |
| 2026-01-08 01:31:28 +0100 | annamalai | (~annamalai@2409:4042:eb8:bd50::9eca:160e) (Ping timeout: 246 seconds) |
| 2026-01-08 01:31:39 +0100 | trickard_ | (~trickard@cpe-50-98-47-163.wireline.com.au) |
| 2026-01-08 01:36:01 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 264 seconds) |
| 2026-01-08 01:36:37 +0100 | shr\ke | (~shrike@user/shrke:31298) (Ping timeout: 264 seconds) |
| 2026-01-08 01:36:55 +0100 | trickard_ | trickard |
| 2026-01-08 01:37:47 +0100 | shr\ke | (~shrike@user/paxhumana) paxhumana |
| 2026-01-08 01:37:47 +0100 | shr\ke | (~shrike@user/paxhumana) (Changing host) |
| 2026-01-08 01:37:47 +0100 | shr\ke | (~shrike@user/shrke:31298) shr\ke |
| 2026-01-08 01:40:21 +0100 | trickard | (~trickard@cpe-50-98-47-163.wireline.com.au) (Read error: Connection reset by peer) |
| 2026-01-08 01:42:14 +0100 | trickard_ | (~trickard@cpe-50-98-47-163.wireline.com.au) |
| 2026-01-08 01:46:42 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 01:49:16 +0100 | raincomplex | (~rain@user/raincomplex) raincomplex |
| 2026-01-08 01:51:15 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) (Ping timeout: 240 seconds) |
| 2026-01-08 01:52:16 +0100 | L29Ah | (~L29Ah@wikipedia/L29Ah) L29Ah |
| 2026-01-08 01:55:04 +0100 | <EvanR> | in the sense that a program is a proof of some theorem, if I can get a computer to give me the correct program that's great. But that's not what we're doing right now with AI |
| 2026-01-08 01:55:48 +0100 | <EvanR> | it's more like mass copywrite infringement search |
| 2026-01-08 01:56:35 +0100 | <jreicher> | Even correct programs aren't proofs of theorems. "In retrospect it seems to be doing the right thing." |
| 2026-01-08 01:56:38 +0100 | <EvanR> | of code that is hardly a proof of anything |
| 2026-01-08 01:56:39 +0100 | ryanbooker | (uid4340@id-4340.hampstead.irccloud.com) ryanbooker |
| 2026-01-08 01:57:22 +0100 | <EvanR> | jreicher, correct in the sense that I explicitly specified the theorem |
| 2026-01-08 01:57:36 +0100 | <EvanR> | not "write me an MMO in haskell" |
| 2026-01-08 01:58:01 +0100 | <newmind> | that would be the dream, yeah: provide a spec, and it spits out a program that fulfills that (and while we're at it, also does proofably terminate and run in limited space).. but current AI agents are doing the exact opposite, little more than running in yolo mode and just executing whatever comes to mind.. what i'm proposing is a middle ground: at |
| 2026-01-08 01:58:02 +0100 | <newmind> | least run code that's checked by a compiler and can't do anything completely nuts |
| 2026-01-08 01:58:48 +0100 | <jreicher> | Even if AI could do it, I think the bigger problem is motivating humans to write specs in the first place. That's been possible for half a century and still almost nobody doesit. |
| 2026-01-08 01:58:49 +0100 | merijn | (~merijn@host-cl.cgnat-g.v4.dfn.nl) merijn |
| 2026-01-08 01:59:15 +0100 | <EvanR> | that part is beyond my paygrade xD |