The nushell pre_prompt hook applied direnv's export diff with `load-env`,
which cannot remove a variable. Variables direnv marks for removal (null
values) were left behind as empty ghosts. Hide those keys explicitly and
drop them before `load-env`.
Add a per-mount 'mountType' option ('mount' | 'nfsmount') selecting
which rclone command performs the mount. 'nfsmount' runs rclone's
in-process NFS server and mounts it via the system NFS client, so no
FUSE driver is needed — on macOS this removes the module's documented
macFUSE requirement (and the kernel-extension approval plus Reduced
Security boot policy that come with it).
When 'nfsmount' is selected, 'nfs-cache-type = "disk"' is merged
into the mount's options (user-overridable, mirroring the existing
'vfs-cache-mode' defaulting) so NFS file handles survive rclone
restarts; with the in-memory default, every service restart would
leave the kernel client with stale handles.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
When services.syncthing.guiAddress is a unix socket, curlAddressArgs built
`--unix-socket <sock> http://.<path>`. curl >= 8.21 tightened URL parsing and
rejects the bare-dot authority with "URL rejected: Bad hostname" (exit 3), so
every REST call in the merge script fails. Combined with `--retry 1000
--retry-delay 1 --retry-all-errors`, syncthing-init.service retries until
systemd terminates it at the start timeout and Home Manager activation fails.
Use `localhost` as the placeholder authority; --unix-socket still routes the
request to the socket, and localhost is accepted by all curl versions. Add a
regression test asserting the unix-socket path uses http://localhost.
User-domain agents are not rediscovered from the user LaunchAgents directory after reboot. Let built-in agents inherit the GUI default while keeping explicit user-domain configuration available.
The mdBook migration split the manual into one HTML page per section,
but same-page fragment links (e.g. [Standalone setup](#sec-flakes-standalone))
were left pointing at anchors that now live on other pages, so they
silently do nothing when clicked.
Rewrite all cross-page fragment links in docs/manual and
docs/release-notes to page-qualified relative links. Option anchor
links (#opt-*, #nixos-opt-*, #nix-darwin-opt-*) are untouched since
convert-markup.py rewrites those at build time.
Also give the previously empty-label [](#sec-tests) link in the 25.05
release notes a "Tests" label, since the mdBook renderer does not
auto-fill labels from the target heading.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
'cfg.diffConfig' is an implicit submodule with sub-options that carry
their own defaults, so it never actually equals `{ }` - the previous
`cfg.diffConfig != { }` check was always true. This caused
'kitty/diff.conf' to be written (and clobber any existing unmanaged file)
even when the user configured nothing under `programs.kitty.diffConfig`.
Add an extraPackages option to the television module that wraps the `tv`
binary with additional packages in PATH. Defaults to fd, bat, and ripgrep
which are the dependencies required by television's default channels.
Existing tests updated to set extraPackages = [ ] since they test config
generation, not binary wrapping.
💘 Generated with Crush
Assisted-by: Crush:glm-5.2
podman pulls in vfkit on darwin, which fails to build on
aarch64-darwin with the current nixpkgs pin (cctools ld crashes
with 'Trace/BPT trap: 5'). The podman darwin tests reference the
real package through the machine watchdog script, establishing a
build dependency. Stub it on darwin test runs.
Both fail to build on aarch64-darwin with the current nixpkgs pin
(cctools ld crashes with 'Trace/BPT trap: 5'): gurk-rs directly, and
vfkit as a colima dependency. Stub them on darwin test runs.
Move public option declarations and file-entry helpers into dedicated files while keeping module assembly in default.nix. Update the AI labeler for the new module directory.
Persist managed plugins in Claude Code's personal skills directory so strict-parser subcommands do not receive injected --plugin-dir arguments. Retain version-gated legacy behavior for older and unversioned packages.
When an agent's launchd.agents.<name>.domain is changed from gui to
user (or the agent is newly installed on a headless macOS system
where the gui/<UID> domain never existed), bootoutAgent attempts to
unload the agent from the old domain first. On such systems, this
fails with: Boot-out failed: 125: Domain does not support specified
action.
The bootoutAgent function only treated "No such process" as a
non-fatal condition (return 2). This change adds the domain-not-exist
error to the same handling path: the agent wasn't running in a domain
that doesn't exist, so we can safely skip the boot-out and proceed
with bootstrap into the new domain.
Follow up on home-cursor: remove legacy enable by preserving the implicit enable path that was not covered by the earlier deprecation warning.
Configurations that set home.pointerCursor without top-level enable now continue to work and emit a warning asking users to set home.pointerCursor.enable explicitly.
yubikey-agent triggers OS-level CCID/PIV PIN entry dialogs when SSH
authentication is requested. These system security prompts require
access to the macOS Aqua session (window server). Running in the user
domain (LimitLoadToSessionType=Background) blocks these prompts,
causing authentication to fail or hang.
Revert to the default gui domain so PIN prompts can render in the
graphical login session.
gpg-agent is commonly configured with graphical pinentry programs
(pinentry-mac, pinentry-gnome3) on macOS that need access to the Aqua
session to display passphrase prompts. Running in the user domain
(LimitLoadToSessionType=Background) prevents these prompts from
rendering, causing passphrase entry to fail silently or hang.
Revert to the default gui domain so the agent runs in the graphical
login session where pinentry can interact with the window server.
Fixes the following accessing issue:
```
Error accessing credential on keyring: PlatformFailure(Error { code: -25308, message: "User interaction is not allowed." })
```
Added a new config option programs.firefox.profiles.<name>.storeId.
Setting this option now writes the StoreID both to the profiles.ini
and to the settings."toolkit.profiles.storeID" of the profile.
Firefox has two different profile implementations;
- the old one that home-manager module can declare
- the new "Selectable Profile Service" introduced in 150
The new profiles are managed in a runtime db, and reads from profiles.ini
Migration without database operations is not possible right now,
but when it becomes available, this option will stop home-manager from
clobbering the linked storeIDs.