| 18 Apr 2022 |
 aanderse | so how does my flake.nix look in your scenario? | 20:45:42 |
| aanderse | Linux Hackerman: yes, that | 20:45:52 |
 Buckley | Probably just use the non-flake hive files in a folder, then they make calls into the top level flake and just select the hosts they want. Although then that would mess with the nodes input… not sure if that’s an issue | 20:47:20 |
| aanderse | definitely need nodes | 20:47:40 |
| Buckley | That’s what I used to do when I used morph anyway | 20:47:44 |
| Buckley | Well then it’ll be a little more involved | 20:48:19 |
| Buckley | You could always procedurally generate your tags… so like produce a tag deploy1_webservers for machines which are both deployment1 and webservers | 20:49:27 |
| Buckley | Not sure exactly what that code would look like right now | 20:49:43 |
 Linux Hackerman | This all sounds like not-so-great workarounds for a problem that would be best solved by adding the feature to colmena. | 20:50:02 |
| Buckley | That would be the over engineered way | 20:50:04 |
| aanderse | ok cool | 20:50:30 |
| Buckley | Tbh I feel like the need for multiple deployments points to a deficcieccy in the tagging system. I actually thought you could combine tags to filter down more | 20:51:06 |
| Linux Hackerman | I don't think so, there are good reasons to keep the deployments separate anyway. Like if the configs aren't supposed to refer to each other. | 20:51:49 |
| aanderse | i have great interest in utilizing nodes
i'm not sure how you would make that work (well) with the tagging system | 20:51:50 |
| aanderse | Linux Hackerman: bingo | 20:51:59 |
| Linux Hackerman | I mean, yes, tag selection could be more powerful and it's something I'd like to have as well. But I'm pretty sure it's not the right solution to aanderse's problem :) | 20:52:23 |
| aanderse | learning rust and contributing a feature might be the solution to my problem... but i feel like that isn't happening soon | 20:53:29 |
| aanderse | so i'll hack something else up for now :) | 20:53:40 |
| 19 Apr 2022 |
 Zhaofeng Li | Sorry all, my homeserver was completely borked for a week and still catching up. The last scrollback I see after April 12 is "Not sure exactly what that code would look like right now" from earlier today. Could someone fill me in with the conversation? | 06:50:19 |
| Linux Hackerman | 
Download 2022-04-19-090015_screenshot.png | 07:00:34 |
| Linux Hackerman | (because screenshots are obviously the best way to share chat transcripts 🙄) | 07:00:49 |
| Zhaofeng Li | Thanks! I can see that part fine, what about the messages before that? Or how should I go about forcing Synapse Web to refetch the history? | 07:03:14 |
| Linux Hackerman | oh oops | 07:03:25 |
| Linux Hackerman | I thought you could only see up to that message 🙃 | 07:03:37 |
| Zhaofeng Li | Okie, "Clear cache and reload" did the trick. Sometimes eventual consistency needs a kick 😅 | 07:05:30 |
| Zhaofeng Li | In reply to @aanderse:nixos.dev
defaults apply to all machines regardless of tags, i don't want to accidentally reference a machine in a different tag via nodes, i want to grab all attributes from all machines in a tag Yeah, nodes refers to all nodes regardless of selection. We could have something like lib-colmena.findNodesByTag "webservers" but not sure whether that's the best way to solve the issue | 07:09:43 |
|  andi- joined the room. | 07:37:06 |
 David Arnold (blaggacao) | Zhaofeng Li: you might have a look at this new exciting project: https://github.com/numtide/nix-eval-cache It uses black magic to cache better than nix flakes. | 16:48:47 |
| David Arnold (blaggacao) | In reply to @zhaofeng:zhaofeng.li
Yeah, nodes refers to all nodes regardless of selection. We could have something like lib-colmena.findNodesByTag "webservers" but not sure whether that's the best way to solve the issue I think the way to go about these issues is to modularize the eval interface. Batteries included, but if your project requires a different eval structure, that's not something that colmena should implement. | 18:55:13 |
| 20 Apr 2022 |
|  rh joined the room. | 23:31:38 |
