19 Sep 2024 |
Artturin | K900: > if it needs changes then let's discuss in matrix.to#/#stdenv:nixos.org lol | 18:11:59 |
Artturin | *
if it needs changes then let's discuss in matrix.to#/#stdenv:nixos.org lol K900
| 18:13:06 |
Artturin | *
if it needs changes then let's discuss in matrix.to#/#stdenv:nixos.org
lol K900
| 18:13:11 |
Artturin | *
if it needs changes then let's discuss in matrix.to#/#stdenv:nixos.org
lol K900
| 18:13:15 |
Artturin | * I said
if it needs changes then let's discuss in matrix.to#/#stdenv:nixos.org
lol K900
| 18:13:34 |
emily | do we want to couple it with a ratchet check (in nixpkgs-vet?) to prevent new occurrences? | 18:14:27 |
emily | or is the idea to just whack-a-mole them a few times before adding a warning? | 18:14:37 |
Artturin | nixpkgs-vevt is just a pkgs/by-name tool, Idk why it's name was changed to that | 18:15:31 |
Artturin | Or has it changed | 18:15:59 |
emily | I think it was changed specifically with an eye towards implementing more ratchet checks | 18:17:20 |
emily | "Tool to vet (check) Nixpkgs, including its pkgs/by-name directory" | 18:17:31 |
Artturin | Doesn't look like it has anything to support this kind of thing yet | 18:19:49 |
Artturin | K900:
Idea is OK, list of merge conflicts is scary. Maybe do this stepwise?
What's the advantage?
| 18:21:25 |
20 Sep 2024 |
Tristan Ross | In reply to @artturin:matrix.org
K900:
Idea is OK, list of merge conflicts is scary. Maybe do this stepwise?
What's the advantage?
Easier to manage, GH restricts the size of the diff. Unless I want to build a PR, I wouldn't want to review in the terminal. GH's diff viewer is pretty convenient. | 02:40:10 |
Artturin | In reply to @rosscomputerguy:matrix.org Easier to manage, GH restricts the size of the diff. Unless I want to build a PR, I wouldn't want to review in the terminal. GH's diff viewer is pretty convenient. There's simply no way to review this change manually, the only way is to copy the commands in the commit msg and diffoscope, but first drop the formatting commit because I didn't post the command for it(I'll add it later) | 02:45:03 |
Artturin | Maybe there's performance issues when rebasing and getting conflicts from a large commit? | 02:47:07 |
Artturin | That would be a valid reason to split it | 02:47:26 |
Artturin | * That would be a valid reason to split the change in to a few commits in the same PR | 02:48:10 |
Tristan Ross | Maybe but still like to review manually | 02:52:18 |
emily | you're going to manually review >4,000 files? | 11:30:51 |
emily | I think spot checks are the most we can hope for here | 11:31:02 |
emily | the important thing is just it getting 0 rebuilds | 11:31:18 |
Artturin | In reply to @emilazy:matrix.org the important thing is just it getting 0 rebuilds https://github.com/NixOS/nixpkgs/pull/341407#issuecomment-2353173805 | 11:32:55 |
emily | right :) | 11:33:19 |
emily | I shouldn't have said 0 strictly | 11:33:31 |
emily | I more meant "if we can account for every single rebuild that gives us stronger confidence we're not changing semantics than manual review could" | 11:33:47 |
| aleksana (force me to bed after 18:00 UTC) joined the room. | 14:13:25 |
| cafkafk joined the room. | 14:29:50 |
Tristan Ross | In reply to @emilazy:matrix.org you're going to manually review >4,000 files? No, I want to review the stuff I am a codeowner/maintainer of. | 16:47:04 |
Artturin | @rosscomputerguy:matrix.org: try files -> file filter -> files owned by you | 17:39:49 |