| 1 Mar 2023 |
@kranzes:matrix.org | Instead of relying on the default/auto options | 00:12:53 |
@lily:lily.flowers | Here's the diff I'm currently testing:
diff --git a/pkgs/os-specific/linux/plymouth/default.nix b/pkgs/os-specific/linux/plymouth/default.nix
index f2abfa68682..42ba23901e0 100644
--- a/pkgs/os-specific/linux/plymouth/default.nix
+++ b/pkgs/os-specific/linux/plymouth/default.nix
@@ -21,6 +21,8 @@ stdenv.mkDerivation rec {
pname = "plymouth";
version = "unstable-2023-02-22";
+ outputs = [ "out" "dev" ];
+
src = fetchFromGitLab {
domain = "gitlab.freedesktop.org";
owner = pname;
@@ -30,7 +32,6 @@ stdenv.mkDerivation rec {
};
patches = [
- ./add-option-for-installation-sysconfdir.patch
./dont-create-broken-symlink.patch
];
@@ -55,7 +56,6 @@ stdenv.mkDerivation rec {
];
mesonFlags = [
- "--sysconfdir=/etc"
"-Dlogo=/etc/plymouth/logo.png"
"-Dbackground-color=0x000000"
"-Dbackground-start-color-stop=0x000000"
@@ -74,6 +74,10 @@ stdenv.mkDerivation rec {
postPatch = ''
substituteInPlace meson.build \
+ --replace "'PLYMOUTH_THEME_PATH', plymouth_theme_path" "'PLYMOUTH_THEME_PATH', '/etc/plymouth/themes'" \
+ --replace "'PLYMOUTH_PLUGIN_PATH', plymouth_plugin_path" "'PLYMOUTH_PLUGIN_PATH', '/etc/plymouth/plugins/'" \
+ --replace "'PLYMOUTH_POLICY_DIR', plymouth_policy_dir" "'PLYMOUTH_POLICY_DIR', '/etc/plymouth/'" \
+ --replace "'PLYMOUTH_CONF_DIR', plymouth_conf_dir" "'PLYMOUTH_CONF_DIR', '/etc/plymouth/'" \
--replace "run_command(['scripts/generate-version.sh'], check: true).stdout().strip()" "'${version}'"
sed -i '/^install_emptydir/d' src/meson.build
| 00:13:17 |
@lily:lily.flowers | I left most of the meson flags as default | 00:13:33 |
@lily:lily.flowers | I'm double checking that flicker-free boot in a VM works seamlessly like before just to be sure there are no regressions from the update | 00:14:42 |
@lily:lily.flowers | In reply to @kranzes:matrix.org Are the unnecessary things other meson flags? (so to clarify, unnecessary things were the sysconf related patch and flag, since it does the right thing by default and has similar semantics to the autoconf script -- which means we still need to manually patch the compiled-in definitions like the autoconf scripts though, it seems) | 00:16:41 |
@kranzes:matrix.org | Can't view diff on mobile oof | 00:18:25 |
@kranzes:matrix.org | Anyway I'm going to bed now, i need to be at school in 6 hours 😬 | 00:19:12 |
@kranzes:matrix.org | I'll check your suggestions tomorrow | 00:19:26 |
@lily:lily.flowers | Sounds good, rest well! | 00:19:33 |
@kranzes:matrix.org | Yeah im really gonna enjoy these 4 or 5 hours haha 🙃 | 00:20:01 |
@kranzes:matrix.org | * Yeah im really gonna enjoy these 4 or 5 hours of sleep haha 🙃 | 00:20:18 |
| 2 Mar 2023 |
| @maxeaubrey:matrix.org changed their display name from maxine (they/them) to maxine. | 00:58:05 |
| 6 Mar 2023 |
@elvishjerricco:matrix.org | alright back to the networkd branch... | 18:57:27 |
@elvishjerricco:matrix.org | That's interesting... I see output from the openvpn command before I see systemd[1]: Starting openvpn.service... | 19:06:07 |
@lily:lily.flowers | In reply to @elvishjerricco:matrix.org That's interesting... I see output from the openvpn command before I see systemd[1]: Starting openvpn.service... Is that in a NixOS test? I've noticed it too before in those and wondered if it had to do with the buffering the serial console grabber has (and it maybe interacting poorly with journald) | 19:14:41 |
@elvishjerricco:matrix.org | yea it's a nixos test | 19:15:02 |
@elvishjerricco:matrix.org | that could be the case | 19:15:15 |
@elvishjerricco:matrix.org | this is all very nearly simultaneous | 19:15:22 |
@elvishjerricco:matrix.org | Well, it appears I've got openvpn started and the ip address is set, but the server can't ping it.. | 20:21:31 |
@elvishjerricco:matrix.org | * Well, it appears I've got openvpn started and the ip address is set, but the server can't ping it | 20:21:34 |
@elvishjerricco:matrix.org | Oh. That might be because the non-systemd-initrd test is also failing to ping it... | 20:24:23 |
@elvishjerricco:matrix.org | Derp. It's because I commented out the minimalboot stuff, which changed the DHCP lease ordering, so the numbers weren't hard coded right | 20:58:31 |
@elvishjerricco:matrix.org | But, now that the test works, openvpn in initrd boots like 6 seconds faster for some reason | 20:58:51 |
@elvishjerricco:matrix.org | Maybe because openvpn can start in parallel with nixos activation? | 20:59:13 |
| 7 Mar 2023 |
@elvishjerricco:matrix.org | So in order to use networkctl, you need dbus, as far as I can tell. | 02:27:53 |
@elvishjerricco:matrix.org | Ugh | 02:27:57 |
@elvishjerricco:matrix.org | But NixOS supports two different dbus implementations so that makes it trickier to implement that | 02:34:23 |
@lily:lily.flowers | In reply to @elvishjerricco:matrix.org But NixOS supports two different dbus implementations so that makes it trickier to implement that I mean, we could technically only support one for initrd, because it doesn't need to live past initrd, right? | 02:37:11 |
@elvishjerricco:matrix.org | Possibly. Just feels a little disingenuous to tell the user they've replaced dbus with broker, but then not do that in initrd :P | 02:37:55 |
@lily:lily.flowers | Yeah, that's fair | 02:38:42 |