fix(cli): resolve paths from THIS install, not from $HOME#8
Open
damonleelcx wants to merge 2 commits into
Open
fix(cli): resolve paths from THIS install, not from $HOME#8damonleelcx wants to merge 2 commits into
damonleelcx wants to merge 2 commits into
Conversation
The installer has supported --install-dir / --sandbox for a long time, but the
CLI had no notion of it: ~24 sites derived their paths from the user's home via
dirs::home_dir().join(".aikey"). A relocated install's `aikey` therefore operated
on the DEFAULT install's vault, config, run state and logs — two installs, one
data plane, no error message anywhere.
resolve_aikey_dir() becomes the single resolver, in this order:
1. AIKEY_HOME, absolute only. A relative value is ignored rather than honoured:
these paths are written into config and handed to services, where a value
that depends on the current working directory is a latent bug (the same
class as the CWD-relative `.aikey/...` collapse that bit aikey-local-server
under systemd). Falling back to the historical default is the safer failure.
2. Derived from the binary's own location: <root>/bin/aikey → <root>. This is
what makes relocation work with nothing exported. AIKEY_HOME covers the
installer's own sub-invocations, but a user typing `aikey` months later has
no such variable — and the binary they just ran already knows which install
it belongs to. Two guards: the parent directory must be named `bin`, so a
dev build under target/debug falls through; and the candidate root must
carry an install marker, so a shim dropped in /usr/local/bin does not make
us treat /usr/local as an install and start writing a vault there.
current_exe() resolves symlinks, so a link in ~/bin still points home.
3. $HOME/.aikey — unchanged for every ordinary install.
All production callers route through it; no module rebuilds the path by hand.
Verified live with the same $HOME for both binaries: the relocated one targets
its own vault (<root>/data/vault.db) while the real install's binary still
targets ~/.aikey/data/vault.db, and a dev build is unaffected. 1138 unit tests
pass, including four new ones covering override, verbatim use, relative-ignored
and empty-value fallback.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
…lows-its-own-install
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
The installer has supported
--install-dir/--sandboxfor a long time, but the CLI had no notion of it: ~24 sites derived their paths from the user's home viadirs::home_dir().join(".aikey"). A relocated install'saikeytherefore operated on the default install's vault, config, run state and logs — two installs, one data plane, and no error message anywhere.resolve_aikey_dir()becomes the single resolver:AIKEY_HOME, absolute only. A relative value is ignored rather than honoured — these paths are written into config and handed to services, where a value that depends on the current working directory is a latent bug (same class as the CWD-relative.aikey/...collapse that bitaikey-local-serverunder systemd). Falling back to the historical default is the safer failure.<root>/bin/aikey→<root>). This is what makes relocation work with nothing exported:AIKEY_HOMEcovers the installer's own sub-invocations, but a user typingaikeymonths later has no such variable — and the binary they ran already knows which install it belongs to.$HOME/.aikey— unchanged for every ordinary install.Two guards on the exe-derivation: the parent directory must be named
bin(so a dev build undertarget/debugfalls through), and the candidate root must carry an install marker (so a shim in/usr/local/bindoesn't make us treat/usr/localas an install and start writing a vault there).current_exe()resolves symlinks, so a link in~/binstill points home.Verification
Live, with the same
$HOMEfor both binaries:<root>/bin/aikey)<root>/data/vault.db~/.aikey/bin/aikey)~/.aikey/data/vault.db✓ foundtarget/release/aikey)~/.aikey/data/vault.db— unaffectedBefore this change all three pointed at the last row.
1138 unit tests pass, including four new ones covering override, verbatim use, relative-ignored, and empty-value fallback.
Notes for review
develop-v1.0.4rather than thefeature/model-mapping-egress-v1.0.4branch this work was written on, so open PR feat(cli): P1e credential re-grain +aikey usetwo-axis summary + provider(s) mislabel fix #6 stays scoped to its own feature. Re-verified on this base.AIKEY_HOMEand relocates the rendered config. Neither depends on the other to be correct, but they are the same defect family.🤖 Generated with Claude Code