Use head_info rather than stacks or stack_details when possible

[Caleb-T-Owens]

We still need some usage of stacks for the branch listing page AFAIK, but using head_info for the main projection is a big help since it avoids the issues with some strange inconsistency issues between stack_details and stacks.

[krlvi]

awesome!

[slarse]

@Caleb-T-Owens Smoke tests failed with

Error: database is locked

Caused by:
    Error code 5: database is locked

That's not related to the code in this PR (see this comment) and probably a spurious error, but it's still worrying. My best guess is that it's the background sync interfering with the next command being executed. but is not great for scripting at the moment.

[Byron]

I think there is a way to turn backtraces on with anyhow, so it can then show the origin of the error. Maybe RUST_BACKTRACE=full` would do the trick? Maybe there is more to it as well, might need some research.

Then once it's clear where this is happening, it can be fixed.
What might be happening here is that something doesn't obtain a repository RW-lock, so the database acts as sync mechanism, and slow-running opts take too long and trigger a busy timeout. Our own repository lock has no timeout.
But… that's just a guess.

[Copilot]

Pull request overview

This PR migrates the desktop app’s primary stack/stack-details projection from legacy stacks/stack_details commands to the newer head_info (RefInfo) API, introducing adapter utilities to map RefInfo into the existing “legacy” stack/branch details shapes used by the UI.

Changes:

• Replaced stacks + stack_details RTK Query endpoints with a single head_info-backed workspaceDetails endpoint and updated StackService consumers accordingly.
• Added headInfoAdapters (and unit tests) to transform RefInfo into WorkspaceDetails (entity state + derived per-stack details).
• Removed legacy Tauri command registration/permissions for stacks and stack_details, and updated PostHog “high volume” command tracking to head_info.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated 5 comments.

Show a summary per file

File	Description
packages/but-sdk/src/test.ts	Updates SDK smoke test to use headInfo + listProjectsStateless.
crates/gitbutler-tauri/src/main.rs	Stops registering stacks/stack_details as invokable Tauri commands.
crates/gitbutler-tauri/permissions/default.toml	Removes stacks/stack_details from the default allowlist.
apps/desktop/src/lib/telemetry/posthog.ts	Treats head_info as the high-volume command.
apps/desktop/src/lib/stacks/stackService.svelte.ts	Switches stack/branch/commit selectors to read from workspaceDetails instead of legacy endpoints.
apps/desktop/src/lib/stacks/stackEndpoints.ts	Replaces legacy endpoints with workspaceDetails query backed by head_info + adapters.
apps/desktop/src/lib/stacks/headInfoAdapters.ts	New adapter layer transforming RefInfo into stack/stack-details entities.
apps/desktop/src/lib/stacks/headInfoAdapters.test.ts	Adds Vitest coverage for the RefInfo → WorkspaceDetails mapping.
apps/desktop/src/components/branchesPage/BranchesViewStack.svelte	Switches branches-page stack lookup from “all stacks” to workspaceDetails stack lookup.

[Copilot]

Pull request overview

Copilot reviewed 14 out of 15 changed files in this pull request and generated 3 comments.

[Caleb-T-Owens]

I've come into the realisation that my mapping layer to the old datastructures introduces a potential regression.

If you have a branch segment that doesn't have a reference, it fails hard.

Anon segments really need to be a first class supported entity, so I'm switching the frontend over to using the new datastructures which will allow us to properly support the scenario.