!9IQChSjwSHXPPWTa:lix.systems

Lix

1106 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-rooms295 Servers

Load older messages


SenderMessageTime
21 Nov 2025
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)ah, yea17:45:11
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her) I mean, there are some issues but they are fixed already
I also once found an binary that every time when u download it, has a different hash
17:46:02
@charles:computer.surgeryCharlesi.e. reproducible build environments are necessary but not sufficient to achieve Reproducible Builds17:46:08
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)is this more like, clangs job now?17:46:33
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)like17:46:41
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her) we have the environment 17:46:45
@charles:computer.surgeryCharlesno, it's the job of every piece of software involved in producing the artifacts17:46:54
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)ah17:47:03
@charles:computer.surgeryCharlessource code, compilers, build scripts, etc; things of that nature17:47:03
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)what would be the most finniky part currently?17:47:25
@charles:computer.surgeryCharles all a nix implementation can do is ensure that the build environment is bit-exact reproducible, not the output artifacts from any given build 17:47:34
@charles:computer.surgeryCharlesbitching at every upstream to write deterministic code for generating outputs of any form17:48:00
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)ooo, 3?17:48:26
@charles:computer.surgeryCharlesgetting the current date? not reproducible. iterating a hashmap? not reproducible. etc.17:48:39
@piegames:flausch.socialpiegames(btw we do have governance since this summer)17:49:45
@522_:catgirl.cloud522 it/its ⛯ΘΔ running builds in an environment that always does stuff like iterate files in alphabetical order would at least work around reproducibility issues 17:51:00
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)https://medium.com/the-toit-take/hash-maps-that-dont-hate-you-1a96150b492a at least hash maps could be done with this but idk17:51:11
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)yea17:51:22
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)oooo17:51:33
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)nice17:51:35
@charles:computer.surgeryCharlesthe point is that this is a problem in general, sure you can solve it in specific cases, but you'd need to convince every upstream to fix this, which is not gonna happen17:51:45
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)has aux gotten their team setup already?17:51:52
@522_:catgirl.cloud522 it/its ⛯ΘΔi mean you can freely use hashmaps that don't maintain insertion order, you just need to not leak that ordering to the end binary17:52:03
@charles:computer.surgeryCharlesyeah you just need to convince every piece of software in your build environment to do this, which is not gonna happen17:52:15
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)I mean, yea17:52:15
@k900:0upti.meK900 r11y is already being actively worked on in nixpkgs 17:52:16
@k900:0upti.meK900 And massively improving 17:52:25
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her)nice17:52:33
@522_:catgirl.cloud522 it/its ⛯ΘΔno i mean disorderfs exists, and you can choose the build path in nix, no? you can run the build in a disorderfs (which is just FUSE) directory set to sort dirents17:53:44
@sofiedotcafe:matrix.orgSofie 🏳️‍⚧️ (she/her) can't you do this for every nix build? implement in lix or something 17:54:16

Show newer messages


Back to Room ListRoom Version: 10