| 20 Jul 2022 |
 infinisil | Regarding GitHub vs Discourse, I think both have their place: GitHub for task tracking and persistent development discussions, while Discourse is better when we need feedback from the wider community and end-users | 09:41:34 |
|  tim joined the room. | 09:54:38 |
| infinisil | Made some proposed adjustments to the main team document, feel free to take a look: https://github.com/nixpkgs-architecture/.github/pull/2 | 10:00:27 |
|  lvkm joined the room. | 10:56:55 |
|  squalus joined the room. | 13:00:09 |
| infinisil | We'll have the second meeting shortly! | 14:53:14 |
|  mkaito joined the room. | 14:58:25 |
| infinisil | @room Next meeting is now in https://meet.jit.si/nixpkgs-architecture if you want to join :) | 15:01:21 |
 Taeer Bar-Yam | Alas, I can't make it today. Next time! | 15:01:57 |
 j-k | same. have a work demo to deliver | 15:02:36 |
| infinisil | Thanks for joining again everybody! Feel free to fix up https://pad.lassul.us/uIi7xeSJTW6LJUEHulZgVQ a bit, I'll only put it into the meetings repository later | 16:05:39 |
 problems | wasm may be an option if there's a sufficiently portable interpreter but i'm not aware of one atm | 16:05:49 |
| problems | it does hard-rely on ieee floating point behavior, so that excludes... vax :p | 16:06:16 |
 yorik.sar | What's "sufficiently portable"? | 16:06:20 |
| infinisil | I guess Nix is mostly used on just a couple platforms, but there's effort by a bunch of people to support more exotic platforms | 16:07:15 |
| yorik.sar | I mean, we should probably define that. | 16:07:16 |
| yorik.sar | There's also https://github.com/SwadicalRag/wasm2lua that could be used as last resort for such platforms... | 16:08:19 |
 kevincox | FWIW I don't think we need to exactly define it. We consider it as part of the tradeoff. If two options are basically the same but one supports 10 "systems" and the other only supports 8 we prefer the 10 one. However there is probably a minimum set that is a deal breaker that we can define. | 16:08:21 |
 Gytis Ivaskevicius | For lua vs ruby argument:
Initially i wrote most of this in lua, but its nowhere near as clean, requires quite a few extra dependencies and overall less convenient
https://github.com/gytis-ivaskevicius/nix-reinventing-the-wheel/tree/master/builders/scripts
And i was learning ruby with this project, there definetelly are some issues, off the top of my head it requires a little DSL for env/json values access and make spawning processes more asynchronous. All in all - till now i cant think of a better language choice while keeping our restrictions in mind | 16:08:43 |
| yorik.sar | Well, we should probably take into account what platforms we want to support in the end... | 16:08:50 |
| infinisil | For some exotic platforms, this user seems to do some work: https://github.com/NixOS/nixpkgs/pulls?q=author%3Aamjoseph-nixpkgs | 16:08:57 |
| yorik.sar | We could also fall back to generating Bash code... | 16:09:48 |
| problems | the point is to do things bash can't do, like working with structured data | 16:10:43 |
| Gytis Ivaskevicius | In reply to @yorik.sar:matrix.org
We could also fall back to generating Bash code... I'd say either we keep mkderivation as it is or change it for the better | 16:10:51 |
| infinisil | ash (it/its) 🏳️⚧️: I think yorik.sar means to essentially compile to bash from the language that's picked | 16:11:03 |
| Gytis Ivaskevicius | In reply to @yorik.sar:matrix.org
We could also fall back to generating Bash code... * I'd say either we keep mkderivation as it is or change it for the better. No middle ground | 16:11:03 |
| infinisil | I think that's a viable temporary solution while the language waits for support for more exotic platforms | 16:11:25 |
| yorik.sar | By the way, it's also an option - use Bash as runtime, but add a good language that can be translated to it. Then the translator is responsible for all type checking and guarnatees. | 16:11:38 |
| infinisil | Oh but then the compiler from language X to bash needs to be portable.. | 16:11:57 |
| infinisil | Sounds messy! | 16:12:09 |
