!9IQChSjwSHXPPWTa:lix.systems

Lix

737 Members
Lix user channel. Feel free to discuss on-topic issues here and give each other help. For matrix.to links to the rest of the Lix channels, see: https://wiki.lix.systems/books/lix-organisation/page/matrix-rooms225 Servers

Load older messages


SenderMessageTime
11 Dec 2024
@capypara:matrix.orgCapyparaIt shouldn't, but I'm using Jovian (NixOS modules and packages to replicate SteamOS), so it could also be a bug there. 17:47:17
@aloisw:julia0815.dealoisw nix-diff can be very useful for investigating this kind of stuff. 17:48:21
@leo60228:matrix.orgleo60228
In reply to @emma:rory.gay
yeahhh no, the diff part is solely a frontend thing, also git's diff format inside of object packs is dense
well you can make a binary format diff using the packfile format with git bundle create out.bundle commit~1..commit
21:18:13
@leo60228:matrix.orgleo60228
In reply to @emma:rory.gay
git diff just uses a local diff util :)
wdym? git diff is its own diff implementation (you can use it standalone with git diff --no-index)
21:19:06
12 Dec 2024
@atagen:imagisphe.reatagenI noticed Lix uses a canary in the overlay to prevent being overlaid twice.. I'm facing something kind of similar with another overlay I'm consuming, where it applies twice if I specify it in different modules I consume and breaks everything - is this just a corner of nix where there be dragons and there is really no other way?02:37:13
@atagen:imagisphe.reatagen * I noticed Lix uses a canary in the overlay to prevent being overlaid twice.. I'm facing something kind of similar with another overlay I'm consuming, where it applies twice if I specify it in each of several modules I consume, breaking everything - is this just a corner of nix where there be dragons and there is really no other way?02:37:39
@piegames:flausch.socialpiegamesYes06:29:39
@joscha:serwm.comJoscha joined the room.08:46:08
13 Dec 2024
@lillecarl:matrix.orgLilleCarl (Salsa9) joined the room.02:04:37
@acidbong:envs.net@acidbong:envs.neti see 2.91 branch hasn't been updated in 2 months, will all the further commits go to 2.92?19:00:15
@piegames:flausch.socialpiegamesYes, that's how the release cycle goes19:09:00
@acidbong:envs.net@acidbong:envs.netepic19:12:47
14 Dec 2024
@jakra:frei.chatjakraXJ1145c12:17:26
@navi:tchncs.demrtz joined the room.12:28:32
@acidbong:envs.net@acidbong:envs.netwut?12:32:54
@antifuchs:asf.computerantifuchsCatlike typing detected14:21:37
@gsaurel:laas.frnim65smaybe a password, but not a very strong one 😅16:25:12
@tebriel:frodux.nettebriel joined the room.17:08:14
@duanin2:duanin2.topDuanin2 joined the room.18:10:24
@adda0:matrix.orgAdda left the room.18:22:01
@adda0:matrix.orgAdda joined the room.18:24:01
@soliprem:soliprem.euSoliprem joined the room.18:42:42
@aftix:matrix.orgaftix lix is adding ?rev= to locked tarball urls in flake.lock, specifically for git.lix.systems. ex: "url": "https://git.lix.systems/api/v1/repos/lix-project/nixos-module/archive/<sha>.tar.gz?rev=<sha>" this only happens with the lix and lix-module input, and my CI on github that's using cppnix 2.24 really does not like it and fails anytime it gets through. I'm on 2.92.0-dev-pre20241211-92ed9fe 20:56:50
@kfears:matrix.orgKFears (annoying)You might be hitting this? https://git.lix.systems/lix-project/lix/issues/47221:53:06
15 Dec 2024
@9hp71n:matrix.orgghpzin Both should resolve to url with ?rev=<sha> because response has header with link and rel="immutable" that looks that way:
curl -s -o /dev/null 'https://git.lix.systems/api/v1/repos/lix-project/nixos-module/archive/c374ebf5548c7
b6d4bf884369a5b6879cfc916ea.tar.gz' --write-out %{header_json} | jq '.link[0]'
"<https://git.lix.systems/api/v1/repos/lix-project/nixos-module/archive/c374ebf5548c7b6d4bf884369a5b6879cfc916ea.tar.gz?rev=c374ebf5548c7b6d4bf884369a5b6879cfc916ea>; rel=\"immutable\""
06:20:58
@9hp71n:matrix.orgghpzin Both should resolve to url with ?rev=<sha> because response has header with link and rel="immutable" that looks that way:
curl -s -o /dev/null 'https://git.lix.systems/api/v1/repos/lix-project/nixos-module/archive/c374ebf5548c7b6d4bf884369a5b6879cfc916ea.tar.gz' --write-out %{header_json} | jq '.link[0]'
"<https://git.lix.systems/api/v1/repos/lix-project/nixos-module/archive/c374ebf5548c7b6d4bf884369a5b6879cfc916ea.tar.gz?rev=c374ebf5548c7b6d4bf884369a5b6879cfc916ea>; rel=\"immutable\""
06:21:21
@n0emis:noemis.meember joined the room.16:08:42
@n0emis:noemis.meember

Hey, I have an issue with nix-store-paths being copied to the nix-store and therefore evals taking much longer.
The issue appears for example if I set nixpkgs in the flake registry:

~ nix shell --log-format internal-json nixpkgs#hello --command hello
@nix {"action":"start","id":211286621159424,"level":4,"parent":0,"text":"evaluating derivation 'flake:nixpkgs#hello'","type":0}
@nix {"action":"start","id":211286621159425,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/stdenv/generic/default-builder.sh' to the store","type":0}
@nix {"action":"stop","id":211286621159425}
@nix {"action":"start","id":211286621159426,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/stdenv/generic/builder.sh' to the store","type":0}
@nix {"action":"stop","id":211286621159426}
@nix {"action":"start","id":211286621159427,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/build-support/setup-hooks/audit-tmpdir.sh' to the store","type":0}
[...]

(output shortened for brevity)

when I don't set the flake registry, this does not happen:

~ nix shell --log-format internal-json nixpkgs#hello --command hello
@nix {"action":"start","id":209233626791936,"level":4,"parent":0,"text":"evaluating derivation 'flake:nixpkgs#hello'","type":0}
@nix {"action":"start","id":209233626791937,"level":4,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source'","type":0}
@nix {"action":"stop","id":209233626791937}
@nix {"action":"stop","id":209233626791936}
@nix {"action":"start","id":209242216726528,"level":6,"parent":0,"text":"querying info about missing paths","type":0}
@nix {"action":"stop","id":209242216726528}
@nix {"action":"start","id":209242216726529,"level":0,"parent":0,"text":"","type":102}
@nix {"action":"start","id":209242216726530,"level":0,"parent":0,"text":"","type":104}
@nix {"action":"start","id":209242216726531,"level":0,"parent":0,"text":"","type":103}
@nix {"action":"start","id":209242216726532,"level":6,"parent":0,"text":"querying info about missing paths","type":0}
@nix {"action":"stop","id":209242216726532}
@nix {"action":"result","fields":[0,0,0,0],"id":209242216726530,"type":105}
@nix {"action":"result","fields":[0,0,0,0],"id":209242216726531,"type":105}
@nix {"action":"result","fields":[101,0],"id":209242216726529,"type":106}
@nix {"action":"result","fields":[100,0],"id":209242216726529,"type":106}
@nix {"action":"stop","id":209242216726531}
@nix {"action":"stop","id":209242216726530}
@nix {"action":"stop","id":209242216726529}
Hello, world!

My first thought was, that it might be this issue (on nix, but the same if-statement is also in lix' codebase), but even then the path written to the registry does not end with -source the issue appears. Does someone have an idea, what the issue might be?

16:22:13
@n0emis:noemis.meember *

Hey, I have an issue with nix-store-paths being copied to the nix-store and therefore evals taking much longer.
The issue appears for example if I set nixpkgs in the flake registry:

~ nix shell --log-format internal-json nixpkgs#hello --command hello
@nix {"action":"start","id":211286621159424,"level":4,"parent":0,"text":"evaluating derivation 'flake:nixpkgs#hello'","type":0}
@nix {"action":"start","id":211286621159425,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/stdenv/generic/default-builder.sh' to the store","type":0}
@nix {"action":"stop","id":211286621159425}
@nix {"action":"start","id":211286621159426,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/stdenv/generic/builder.sh' to the store","type":0}
@nix {"action":"stop","id":211286621159426}
@nix {"action":"start","id":211286621159427,"level":5,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source/pkgs/build-support/setup-hooks/audit-tmpdir.sh' to the store","type":0}
[...]

(output shortened for brevity)

when I don't set the flake registry, this does not happen:

~ nix shell --log-format internal-json nixpkgs#hello --command hello
@nix {"action":"start","id":209233626791936,"level":4,"parent":0,"text":"evaluating derivation 'flake:nixpkgs#hello'","type":0}
@nix {"action":"start","id":209233626791937,"level":4,"parent":0,"text":"copying '/nix/store/rhn4fkh7hqjlqs9j6naaq623qpl62yz0-source'","type":0}
@nix {"action":"start","id":209242216726532,"level":6,"parent":0,"text":"querying info about missing paths","type":0}
@nix {"action":"result","fields":[100,0],"id":209242216726529,"type":106}
@nix {"action":"stop","id":209242216726529}
Hello, world!

(also shortened, but only removed some duplicate messages)

My first thought was, that it might be this issue (on nix, but the same if-statement is also in lix' codebase), but even then the path written to the registry does not end with -source the issue appears. Does someone have an idea, what the issue might be?

16:23:08
@n0emis:noemis.meemberI'm running lix 2.91.116:23:39

Show newer messages


Back to Room ListRoom Version: 10