| 4 Jan 2025 |
 ElvishJerricco | oh no I forgot I had more nitpicks but that's all done now. Thanks! | 16:51:07 |
 antifuchs | Hahaha, I’m not sure it’s 100% the right solution, pretty sure that things still get restarted inconveniently even with the patch; but I haven’t tried it lately | 16:52:56 |
| ElvishJerricco | antifuchs: I think it solves the biggest issue, which is unduly stopping systemd-udevd and systemd-networkd when they should instead be restarted. | 16:59:18 |
| antifuchs | Well, it’ll get exercised a bunch very soon ((: | 17:00:02 |
| antifuchs | That reminds me, should stc(-ng) maybe consider newly created user units? I’ve had this happen a few times where rolling out a config with user services wouldn’t start them until a reboot or manual start | 17:01:12 |
| ElvishJerricco | user units aren't handled very much at all currently, IIRC | 17:03:27 |
| ElvishJerricco | might be nice to do something better there | 17:03:37 |
| antifuchs | If there’s no major allergy to handling them, I’ll see about adding something for them eventually | 17:13:46 |
|  palo joined the room. | 21:25:27 |
| 5 Jan 2025 |
| ElvishJerricco | How would we feel about not stripping binaries in systemd stage 1? https://github.com/NixOS/nixpkgs/issues/368249#issuecomment-2571777239 | 22:50:16 |
| ElvishJerricco | I actually don't really understand how stripping is like... a good thing? | 22:52:40 |
| ElvishJerricco | I mean I know it shrinks initrd a little but I don't understand object files enough to understand why we're allowed to just delete all symbols | 22:53:43 |
| 6 Jan 2025 |
 raitobezarius | relocation information is not required when the executable is linked | 02:56:08 |
| raitobezarius | relocation information is only required when you are using it as an object file to relink the library into a new binary | 02:56:19 |
| raitobezarius | that's why deleting all symbols and relocation information cause no issue | 02:56:25 |
| ElvishJerricco | ahh thanks | 03:01:01 |
| ElvishJerricco | still, dunno if it's worth 1MiB of initrd size for something that's potentially buggy? | 03:01:57 |
| ElvishJerricco | though I wonder why it's buggy. Why does stripping uncompressed kernel modules cause problems... | 03:03:34 |
| raitobezarius | My intuition is that kernel modules are not your average ELF and contains non trivial sections that can confuse a dumb binutils strip | 03:23:40 |
| raitobezarius | Compression just makes them not ELF so fixed | 03:23:54 |
| raitobezarius | In reply to @elvishjerricco:matrix.org
still, dunno if it's worth 1MiB of initrd size for something that's potentially buggy? Given that nixpkgs is definitely not going to cater to the embedded ecosystem, I do not see any need to torture oneself over it | 03:24:59 |
 Arian | Embedded folks are gonna be compiling from source anyway most of the time | 07:37:22 |
| raitobezarius | I wish I didn't | 09:25:11 |
| raitobezarius | It's a waste of CPU time | 09:25:16 |
| Arian |
* A new set of system/service credentials are added:
shell.prompt.prefix, shell.prompt.suffix and shell.welcome. At login
time these are propagated into the $SHELL_PROMPT_PREFIX,
$SHELL_PROMPT_SUFFIX, $SHELL_PROMPT_WELCOME environment
variables. These in turn are included in the shell prompt of
interactive shells and shown at login time, via
/etc/profile.d/70-systemd-shell-extra.sh. This functionality is
useful to visually highlight the fact a specific shell prompt
originates from a specific system, execution context or tool. These
credentials and environment variables are supposed to be generically
useful within and outside of the immediate systemd context. It is
also used by 'run0', see above.
| 12:13:49 |
| Arian | neat | 12:13:52 |
| Arian | are we using this already? | 12:14:20 |
|  Frédéric Christ changed their display name from Frédéric Christ 🎄23.12. - 07.01. to Frédéric Christ. | 12:31:49 |
 @jeroen:simonetti.nl | Are there any plans to incorporate NFTSet with the nixos firewall? | 16:40:54 |
 K900 | We already have an nftables-based firewall | 16:41:39 |
