| 11 Dec 2024 |
 puck | yeah, i suspect that got forgotten | 13:12:26 |
 Yureka (she/her) | so... this works as a quick fix for me
diff --git a/lix/libstore/build/local-derivation-goal.cc b/lix/libstore/build/local-derivation-goal.cc
index 794b85031..b136ef957 100644
--- a/lix/libstore/build/local-derivation-goal.cc
+++ b/lix/libstore/build/local-derivation-goal.cc
@@ -1528,7 +1528,7 @@ void LocalDerivationGoal::runChild()
if (i.second.source == "/proc") continue; // backwards compatibility
#if HAVE_EMBEDDED_SANDBOX_SHELL
- if (i.second.source == "__embedded_sandbox_shell__") {
+ if (i.first == "/bin/sh") {
static unsigned char sh[] = {
#include "embedded-sandbox-shell.gen.hh"
};
| 13:47:20 |
| Yureka (she/her) | but the proper fix is setting SANDBOX_SHELL correctly, but I'm not sure how/where to do that properly | 13:48:00 |
| puck | in lix/libstore/meson.build; probably merging the if enable_embedded_sandbox_shell and if busybox.found() things a bit | 14:08:15 |
 piegames | About the tests, is there any good way to get the error messages pushed through CI so that we don't have to dig through build logs anymore? | 14:43:25 |
| piegames | I know that GitLab for example uses some bespoke XML format stolen from JUNIT, but once you speak it the errors get integrated nicely | 14:43:58 |
| puck | it'd be doable to hook this up to do buildbot html output; gerrit also has robot comments which might function for this, but i haven't really tried looking at how this integration would look yet | 14:49:18 |
| Yureka (she/her) | In reply to @puck:puck.moe
in lix/libstore/meson.build; probably merging the if enable_embedded_sandbox_shell and if busybox.found() things a bit https://gerrit.lix.systems/c/lix/+/2305 | 15:49:30 |
| puck | yeah, that seems reasonable to me, but i can't stare at it too hard rn | 15:52:12 |
 9999years | https://gerrit.lix.systems/q/author:lix@piegames.de | 17:42:37 |
| 9999years | why doesn't this show any cls | 17:42:40 |
| 9999years | gerrit is a mystery sometimes | 17:42:45 |
| 9999years | not blocking. it's just a matter of taste & getting repeatedly burned by yaml in various forms | 17:43:24 |
| 9999years | i needed owner:lix@piegames.de. what even is the author field.... | 17:43:51 |
| puck | author is the author of the commit | 17:44:07 |
| puck | and i think piegames' commits are on a different email that gerrit isn't aware of | 17:44:24 |
| 9999years | ohhhh yeah they're from git@piegames.de | 17:44:36 |
| puck | yeah, author:git@piegames.de works | 17:44:40 |
| puck | (author and committer here are the git concept; if you rebase/etc you become committer but author stays the same, etc) | 17:44:57 |
| 9999years | aside: this is about https://gerrit.lix.systems/c/lix/+/2190/comment/e5501eaa_12118801/ | 17:45:47 |
| 9999years | i have had many bad experiences with yaml, i'm not sure this justifies pulling in a yaml parser/formatter (even if we already had one in closure somewhere), i'm not convinced that making these json would have much impact, etc. etc. | 17:46:52 |
| 9999years | so i would certainly like to have the json output be tested in its json form and not after conversion to yaml, but Whatever | 17:47:08 |
| puck | yeah, i'd say it's not going to be the most readable anyways | 17:47:50 |
| puck | i've used yaml in https://puck.moe/git/nixtest partially for output-diff-reasons | 17:48:18 |
| puck | * i've used yaml in https://puck.moe/git/nixtest partially for output-diff-reasons, and you're not supposed to manually write the input for it | 17:49:54 |
| puck | * i've used yaml in https://puck.moe/git/nixtest partially for output-diff-reasons, and you're not supposed to manually write it anyways | 17:49:58 |
 V. 🏳️⚧️ | Redacted or Malformed Event | 18:52:19 |
| piegames | In reply to @9999years:matrix.org
so i would certainly like to have the json output be tested in its json form and not after conversion to yaml, but Whatever I mean the yaml is not getting tested in any besides "does not change" and "a human looks at the diff" | 19:25:38 |
|  rhelmot joined the room. | 19:42:47 |
| rhelmot | I've been having this problem with nix forever and just working around it but now that lix is here and I can have nice things I figured I would ask - does anyone know why the nix/lix testcases fail when you build it in a sandboxed nix daemon that is NOT running on nixos? Asking here first on the off chance that this is a known quantity and I can save myself some debugging time :) | 19:56:53 |
