| 4 Nov 2025 |
 lovesegfault | But did not get it | 17:36:51 |
| lovesegfault | I'll do a pass of the changes since 2.32 and try to remedy this, but maybe we ought to have some kind of CI check for this | 17:37:32 |
| lovesegfault | One of those where you can post a comment to disable the behavior whenever work actually doesn't need rl-notes | 17:37:55 |
 Robert Hensing (roberth) | got annoyed with "darwin" linux-builder today; opened https://github.com/NixOS/nix/pull/14475 | 21:29:50 |
| Robert Hensing (roberth) | I don't know if anyone feels very attached to direct file system access, but it seems like a bad default choice, and not something that's even used very often (?) | 21:30:29 |
| Robert Hensing (roberth) | anyway, making such a change early on in the 26.05 cycle seems like a good idea if you want to | 21:31:04 |
| Robert Hensing (roberth) | I don't think it improves linux-builder anytime soon, but it seems like a nice cleanup to me | 21:31:30 |
| Robert Hensing (roberth) | John Ericson: any reason not to merge https://github.com/NixOS/nix/pull/14469#pullrequestreview-3417559764? | 21:37:20 |
|  ncfavier changed their profile picture. | 23:55:00 |
| 5 Nov 2025 |
| lovesegfault | Robert Hensing (roberth): I was thinking about the ulimit stuff. What if we added a test envvar that makes nix hard-error when it fails to set ulimit, and then used that for the tests that currently fail without it to be skipped? | 01:44:23 |
| Robert Hensing (roberth) | In this case, such a solution would have caused all macOS users to silently pass all the language tests. Not great DX | 10:28:58 |
 John Ericson | In reply to @roberthensing:matrix.org
John Ericson: any reason not to merge https://github.com/NixOS/nix/pull/14469#pullrequestreview-3417559764? No reason, but it's failing CI right now so we can't, right? | 14:23:16 |
| Robert Hensing (roberth) | John Ericson: I figured I'd hang out, https://jitsi.lassul.us/nix-maintainers | 20:08:36 |
| Robert Hensing (roberth) | mschwaig is also there! | 20:08:49 |
 tomberek | fzakaria: Can you re-structure the Kaitai checks such that it is not included into the default devshell? | 20:49:28 |
 fzakaria | can you clarify & why ? | 22:34:58 |
| fzakaria | I structured it like the JSON-schema checks | 22:35:07 |
| fzakaria | (just want to make sure I understand the change to be done) | 22:36:03 |
| tomberek | The desired outcome is that the default devShell does not require the JDK. | 23:13:07 |
| 6 Nov 2025 |
| Robert Hensing (roberth) | Ok to migrate most error trace tests to functional characterization tests? https://github.com/NixOS/nix/issues/14500 | 23:46:36 |
| 7 Nov 2025 |
|  @joshuachp:matrix.org left the room. | 08:35:23 |
 Eelco | Hm, what happened with debug info in the Nix dev shell?
$ gdb --args nix
(gdb) b main
Breakpoint 1 at 0x380090
(gdb) r
(gdb) bt
#0 0x0000000000380090 in main ()
whereas on 2.32 we get:
(gdb) b main
Breakpoint 1 at 0x385040: file ../src/nix/main.cc, line 572.
(gdb) r
(gdb) bt
#0 main (argc=1, argv=0x7fffffff2598) at ../src/nix/main.cc:572
| 14:28:33 |
|  @kdeuser:matrix.org left the room. | 17:47:06 |
|  @awwpotato:envs.net changed their display name from awwpotato (she/her) to -> @da157:catgirl.cloud. | 18:30:45 |
| @awwpotato:envs.net left the room. | 18:40:21 |
 Mic92 | Seems to be in gcc only? nix-util-env % file src/libutil/libnixutil.so.2.33.0.p/hash.cc.o
src/libutil/libnixutil.so.2.33.0.p/hash.cc.o: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), with debug_info, not stripped
using nix develop .#native-clangStdenv | 18:58:58 |
| Mic92 | seems fine but with gcc stdenv it's no debug symbols | 18:59:14 |
| Mic92 | use flake .#native-stdenv also seems to be fine | 19:00:28 |
| Mic92 | mhm. normal stdenv is also fine | 19:01:48 |
 Sergei Zimmerman (xokdvium) | separateDebugInfo is on regardless | 19:04:24 |
