!djTaTBQyWEPRQxrPTb:nixos.org

Nixpkgs Architecture Team

224 Members
https://github.com/nixpkgs-architecture, weekly public meetings on Wednesday 15:00-16:00 UTC at https://meet.jit.si/nixpkgs-architecture53 Servers

Load older messages


SenderMessageTime
17 Mar 2024
@infinisil:matrix.orginfinisilhttps://crates.io/crates/insta could do that I believe15:00:25
@infinisil:matrix.orginfinisil(well and running individual tests, kind of related)15:00:53
@willbush:matrix.orgwillbush Ok no problem! I was just wanted to go through each test to make sure I understand why they should pass or fail. However, I can cargo test -- --nocapture and throw some println statements in the cargo tests. 15:01:13
@infinisil:matrix.orginfinisilOh neat, didn't know about that!15:07:48
@infinisil:matrix.orginfinisilOh neat, yes!15:10:06
@tomberek:matrix.orgtomberekFrom playing with this idea before, I really recommend having a separation between the kinds of pure/impure, eval/build, snapshots/historical. Mixing them leads to problems each time.15:22:55
@growpotkin:matrix.orgGrowpotkinI'd echo what Tom said. 15:42:29
@growpotkin:matrix.orgGrowpotkin* I'll echo what Tom said. 15:42:59
@roberthensing:matrix.orgRobert Hensing (roberth)As long as Nixpkgs and NixOS don't start to depend on such a repo in a way that makes using branches and forks harder, why not.18:09:40
@jade_:matrix.org@jade_:matrix.org
In reply to @infinisil:matrix.org
Alternatively, consider a more powerful idea of having a meta-Nixpkgs repository, which contains all historical versions of Nixpkgs that were built by Hydra, with all their metadata. This could include things like whether builds succeeded, which versions are available in which channel, what is the latest release, etc.
yes yes yes please do it with clever merge commits on a branch
18:28:40
@ss:someonex.netSomeoneSerge (back on matrix) I presume there would've already been some discussions on this matter and somebody in this room would know where to redirect me: https://github.com/NixOS/nixpkgs/issues/296544 18:34:23
@qyriad:katesiria.orgQyriad
In reply to @jade_:matrix.org
yes yes yes please do it with clever merge commits on a branch
that over git notes or something?
18:35:04
@jade_:matrix.org@jade_:matrix.org
In reply to @qyriad:katesiria.org
that over git notes or something?
maybe also viable?
18:35:31
@jade_:matrix.org@jade_:matrix.org but the thing about funny merge commits is that they work natively in bisect 18:35:53
@qyriad:katesiria.orgQyriad ahh good point  18:36:19
@infinisil:matrix.orginfinisil
In reply to @jade_:matrix.org
yes yes yes please do it with clever merge commits on a branch
Not sure what kind of merge commits you're thinking of
18:40:46
@jade_:matrix.org@jade_:matrix.org
In reply to @infinisil:matrix.org
Not sure what kind of merge commits you're thinking of
merge commits (one parent is the last Hydra run, the other is the current commit of the repo, and you take the tree directly from the current commit)
18:41:53
@jade_:matrix.org@jade_:matrix.org merge strategy: nothing. it's the identical tree as upstream.  18:42:23
@infinisil:matrix.orginfinisilAhh, so there's a parent chain where each commit is a Hydra eval18:42:40
@jade_:matrix.org@jade_:matrix.org
In reply to @infinisil:matrix.org
Ahh, so there's a parent chain where each commit is a Hydra eval
yeah
18:42:44
@jade_:matrix.org@jade_:matrix.org that's the idea 18:42:49
@jade_:matrix.org@jade_:matrix.org i bet this would be a reasonable replacement for usage for channel branches honestly 18:43:21
@jade_:matrix.org@jade_:matrix.org native git and pretty clean solution 18:43:36
@infinisil:matrix.orginfinisilI agree this would solve the channel history problem nicely, an alternative to https://github.com/NixOS/nixpkgs/pull/252057. (though it doesn't solve the channel post-processing problem)18:45:30
@qyriad:matrix.org@qyriad:matrix.org left the room.20:30:21
18 Mar 2024
@jade_:matrix.org@jade_:matrix.org
In reply to @infinisil:matrix.org
I agree this would solve the channel history problem nicely, an alternative to https://github.com/NixOS/nixpkgs/pull/252057. (though it doesn't solve the channel post-processing problem)
hmmm. one could actually do that as well, by merging in a third fake commit per iteration, with the metadata. 
04:02:52
@jade_:matrix.org@jade_:matrix.org jade might be terminally git brained though 04:03:12
@fgaz:matrix.orgfgaz Is there a way to subscribe to new issues tagged architecture? Before the migration to nixpkgs I could just subscribe to the -architecture repo 11:22:41
@infinisil:matrix.orginfinisil fgaz: Afraid not :( 17:44:32
@jade_:matrix.org@jade_:matrix.org
In reply to @fgaz:matrix.org
Is there a way to subscribe to new issues tagged architecture? Before the migration to nixpkgs I could just subscribe to the -architecture repo
adding to my mental list of things i want to attach to repo event hooks. i wish someone would go build this system ;;
18:29:19

Show newer messages


Back to Room ListRoom Version: 9