| 12 Aug 2021 |
 @timdeh:matrix.org | Indeed. I think if something like dhall could relax it's typing, perhaps only under certain conditions, it would actually be ideal. Since alot of the hard to debug infinite recursion errors would just be impossible | 23:34:51 |
 ultranix | now that flakes have become easy enough for me, next understandings will be for the CAS derivations | 23:35:52 |
| @timdeh:matrix.org | Yes. I'm just waiting for the caches to build up a bit so everything is so daunting | 23:37:14 |
| @timdeh:matrix.org | * Yes. I'm just waiting for the caches to build up a bit so everything isn't so daunting. In terms of build times I mean | 23:37:31 |
| @timdeh:matrix.org | fortunately, with RFC 96, we have a solid plan to introduce experimental features by default in a given binary, so we could get CAS to a point of readiness, and then build a binary for it. And then, when it's easily distributable, the caches will increase exponentially | 23:38:23 |
| @timdeh:matrix.org | I also think Trustix and Cache would be a damn near perfect software distribution system, so I've purchased a lot of disk space and vps, and now I just need to find the time 🤣 | 23:46:27 |
| @timdeh:matrix.org | * I also think Trustix and CAS would be a damn near perfect software distribution system, so I've purchased a lot of disk space and vps, and now I just need to find the time 🤣 | 23:47:07 |
 @kraftnix:matrix.org | Trustix sounds so great to me as well, I have a build machine and lots of disk so when I get around to it (and hopefully when someone documents a setup to make it easier for me) I will definitely run a trustix cache | 23:47:43 |
| @kraftnix:matrix.org | It really does solve one of the main issues I have with binary cache approaches, trust. | 23:48:32 |
| @timdeh:matrix.org | well said 😃 | 23:49:49 |
| @kraftnix:matrix.org | cross validation on builds + distributed storage/sharing, trustix + ipfs/ca filestores can't come fast enough :) | 23:50:34 |
| @timdeh:matrix.org | yeah it's lack of documentation that is holding me back as well. I'd probably have to get an understanding of the Trustix types to really get it going. But I've been practicing my Rust 💪 | 23:51:03 |
| @timdeh:matrix.org | yeah, that ipfs bit especially. I was very skeptical of filecoin in the beginning because performance was abismal, but recent iterations had me impressed | 23:52:31 |
| @timdeh:matrix.org | * yeah, that ipfs bit especially. I was very skeptical of filecoin in the beginning because performance was abysmal, but recent iterations had me impressed | 23:53:31 |
| @kraftnix:matrix.org | i'm definitely on the fence around incentivized distributed filesystems (ala filecoin), i think getting the incentives right for that is hard for it not to create an unstable system, we've had torrents/p2p transfers for a while where communities want that, i just think it needs to get easier | 23:54:49 |
| @kraftnix:matrix.org | i think i'm mostly on the fence because i would never put valuable data there because i wouldn't trust the system to be around in many years to come, given the prior i would use it for things like the nix cache, or other media stores, but then why even have an incentivized system for that? i think if people want to store those files they will, we just need to make it easier to share, which doesn't require a blockchain/coin imo. \rantover we should get a devos offtopic | 23:56:59 |
| @kraftnix:matrix.org | * i think i'm mostly on the fence because i would never put valuable data there because i wouldn't trust the system to be around in many years to come, given that prior i would use it for things like the nix cache, or other media stores, but then why even have an incentivized system for that? i think if people want to store those files they will, we just need to make it easier to share, which doesn't require a blockchain/coin imo. \rantover we should get a devos offtopic | 23:57:32 |
| @kraftnix:matrix.org | * i think i'm mostly on the fence because i would never put valuable data there because i wouldn't trust the system to be around in many years to come, given that prior i would use it for things like the nix cache, or other media stores, but then why even have an incentivized system for that? i think if people want to store those files they will, we just need to make it easier to share, which doesn't require a blockchain/coin imo. \rantover we should get a devos offtopic channel | 23:57:53 |
| 13 Aug 2021 |
| @timdeh:matrix.org | Yeah, but it's nice for ephemeral things like software artifacts. Not every cache has to be a genuine software archive like cache.nixos.org | 00:09:20 |
| @kraftnix:matrix.org | also back on topic for devos. is there a way to hook into digga's allProfilesTestFor tests, essentially many of mine don't really pass because profiles which rely on a specific filesystem layout (the test clears previous filesystems so that a host is bootable), so id like to include a specific provision script in the VM (I have this for my custom tests) in the allProfilesTestFor or be able to disable them so I can not run lot's of failing VM tests and just run my custom ones (that provision correctly for the test). | 00:12:47 |
| @timdeh:matrix.org | I know there is a way to disable them | 00:13:19 |
| @kraftnix:matrix.org | the allProfilesTests works nicely for simple laptop configurations but none of my fancier setups / VMs that bring in extra disks | 00:14:03 |
| @timdeh:matrix.org | yeah the allProfilesTests would probably be better suited as an individual test per profile. I think I was just being lazy 😃 | 00:32:23 |
| @kraftnix:matrix.org | Is that now how it used to be before digga? | 00:33:00 |
| @kraftnix:matrix.org | I also remember having issues back then because lots of my profiles were designed not to work in isolation but only when combined in a suite :) although it's possible I was introducing test errors in another way. | 00:33:47 |
| @kraftnix:matrix.org | but since rakeLeaves happened I feel profiles/suites are on a similar UX footing so it's not really like that anymore, rakeLeaves actually caused me to push a lot of logic / multi-profile setups out of suites | 00:35:25 |
| @timdeh:matrix.org | I intended to eventually refactor it into that before it was moved into digga, but I ran out of free time | 00:38:12 |
| @timdeh:matrix.org | and I am quickly running out of excuses as well 😉 | 00:39:08 |
| ultranix | https://github.com/colemickens/nixos-flake-example#warning i like colemickens flake warnings | 00:42:51 |
 @gtrunsec:matrix.org | In reply to @kraftnix:matrix.org
but since rakeLeaves happened I feel profiles/suites are on a similar UX footing so it's not really like that anymore, rakeLeaves actually caused me to push a lot of logic / multi-profile setups out of suites Right, I have the same problem as I mentioned earlier | 00:43:22 |
