| 20 Sep 2023 |
 K900 ⚡️ | Wrote up a full repro case: https://github.com/NixOS/patchelf/issues/520 | 12:08:03 |
| K900 ⚡️ | @nbp how hard would it be to make a build that's exactly like an official nightly build, but without the relrhack? | 12:10:23 |
| K900 ⚡️ | Is there like a Docker container or something? | 12:10:34 |
 nbp | using Mozilla infra, I suppose this might be easy as soon as you tell me which flags should be set/removed. | 12:11:13 |
| K900 ⚡️ | Just reverting https://hg.mozilla.org/mozilla-central/rev/032b87ff55061bcbdc7a85d9e18fde814797073a | 12:11:41 |
| nbp | I can try to push that to our try server. | 12:12:24 |
| nbp | but from what I understand that would use elfhack instead. | 12:12:48 |
| K900 ⚡️ | Well whatever it was using before worked | 12:14:01 |
| K900 ⚡️ | I mostly just want to make sure this is actually the commit that's causing the regression | 12:14:15 |
| K900 ⚡️ | And the only way I can think of doing that is to build with it reverted and see what happens | 12:14:25 |
| K900 ⚡️ | And I don't want to spend an hour waiting for it to build with nixpkgs tools just to find out it was never broken in nixpkgs to begin with | 12:17:25 |
| nbp | a tarball would be enough? | 12:17:28 |
| K900 ⚡️ | For the build artifacts? Yeah | 12:19:54 |
| K900 ⚡️ | Any .so file should be enough by itself it seems | 12:20:13 |
| nbp | estimates: Should take about 1:15:59 (Finished around 2023-09-20 06:40) | 12:25:08 |
| nbp | (assuming the try server answers) | 12:26:13 |
| K900 ⚡️ | Oh | 12:33:02 |
| K900 ⚡️ | I just realized what you're doing lol | 12:33:05 |
| K900 ⚡️ | Thanks :) | 12:33:07 |
 colemickens | I can revert the updates until it gets to a working build and merge the test PR. I'm torn between wanting to understand why the audio tests were broken or just leaving the test PR as-is and accepting it. | 13:15:51 |
| colemickens | I also feel like it might be worth having some mechanism for warning users if we haven't updated in a few days. I'm not sure of the security implications of stale nightly versions. Maybe it's overkill. | 13:16:36 |
| nbp | K900 ⚡️: https://treeherder.mozilla.org/jobs?repo=try&revision=689e449ae03cc00f00a2cf09c8887547fadb644b&selectedTaskRun=MUzH2o8jSIyJOcLgfl3GDw.0 Click on one of the "B" or "Bpgo" and then on the bottom, select the "Artifacts and Debugging Tools", which would list a bunch of things, from which you want the "target.tar.bz2" | 13:25:52 |
| K900 ⚡️ | Thanks | 13:41:59 |
| K900 ⚡️ | Let me try it | 13:42:01 |
| K900 ⚡️ | Yep | 13:43:54 |
| K900 ⚡️ | This one works | 13:43:56 |
| K900 ⚡️ | As in I can patchelf it and then run it | 13:45:16 |
| K900 ⚡️ | So I guess we are pretty confident in the culprit now | 13:45:46 |
| colemickens | I'm confused, I was trying to find last known good commit and now they all seem to work for me? | 14:13:40 |
| colemickens | git remote update; git reset --hard nix-community/master; nix build .; ./result/bin/firefox and it launches fine. | 14:16:16 |
