android: scaffold the Tauri mobile lane and build a debug APK in CI
Android (Tauri) / Android APK (debug) (push) Failing after 21s
Desktop (Tauri) / Windows installer (cross-compiled) (push) Successful in 2m38s
Desktop (Tauri) / Tauri desktop (Linux) (push) Successful in 4m26s
Desktop (Tauri) / Update manifest (push) Successful in 5s

The phone client is Tauri v2 mobile (operator decision), so it reuses the Vue
frontend and the Rust store and sync engine that already exist rather than
becoming a third implementation to keep in step by hand.

gen/android is committed. tauri android init generated it, its own .gitignore
already excludes the build outputs and every keystore file, and CI must not have
to regenerate a project that manifest edits will accumulate in.

What the scaffold confirms is that the image's JDK pin was load-bearing rather
than incidental: Tauri templated Gradle 8.14.3 with AGP 8.11.0, and CI-android's
versions.env records that JDK 25 needs Gradle 9.1.0+ and that anything older
fails with an opaque "25.0.3" message. Picking 17 for ci-tauri-android avoided
exactly that. namespace and applicationId came out as com.fabledsword.thoughtsync,
matching the desktop identifier, so the app-data story stays consistent.

The lane builds a DEBUG APK for arm64 only. Release APKs need signing, and the
keystore has to be generated by the operator and never pass through CI logs or an
agent session — the constraint recorded for the updater key applies unchanged.
Gradle's throwaway debug keystore needs nothing from anyone, so this can prove the
app compiles and packages today and grow a signed job when a key exists. arm64 is
every real device; the image carries the other three ABIs, so widening is a word.

Triggered by frontend/** as well as desktop/**, because generate_context! compiles
the frontend into the app — the same reasoning that widened desktop.yml. Android,
desktop and web are peers on one quality bar, and a frontend commit that skipped
this lane would ship a stale phone build.

Green here will mean it BUILT. A Linux runner cannot execute an APK, so nothing in
this lane proves the app runs, renders, or is usable by finger.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
This commit is contained in:
2026-08-16 23:01:32 -04:00
co-authored by Claude Opus 5
parent be0eb94225
commit 1f140c7457
43 changed files with 1065 additions and 1 deletions
+36
View File
@@ -159,6 +159,42 @@ backend/frontend push.
- No Postgres lane (unchanged): the desktop app's local store + sync behavior is
verified on the operator's machine, not in CI.
## Android (Tauri mobile) lane — `android.yml`, third image
Android is a peer surface, not a desktop variant, so it gets its own workflow.
- **Image:** `git.fabledsword.com/bvandeusen/ci-tauri-android:1.97` (Rust + the
four Android ABIs + Android SDK/NDK + JDK 17 + Node + tauri-cli). Pins held in
lockstep with `ci-tauri` and `ci-tauri-win` — all three compile the same crate.
- **Why a third image.** `ci-android` says in its own header that it excludes the
NDK ("No Flutter, no NDK, no CMake") and carries no Rust; `ci-tauri` has Rust
but no SDK, plus a WebKitGTK stack Android never touches. Tauri needs the NDK
precisely because the core is Rust compiled to Android ABIs.
- **JDK 17, not the 25 `ci-android`/`ci-flutter` ship.** `tauri android init`
GENERATES the Gradle project, so its version isn't ours to choose — it produced
Gradle **8.14.3** + AGP **8.11.0** here. `CI-android/versions.env` records that
JDK 25 needs Gradle 9.1.0+ and that anything older fails with an opaque
`"25.0.3"` message. Bumping this JDK without checking what Tauri templates will
break the build with an error that names nothing useful.
- **`gen/android/` is COMMITTED.** `tauri android init` generates it, and the only
ignores are its own (build outputs, `local.properties`, `key.properties`,
`keystore.properties`). CI must not have to regenerate it, and manifest or
Gradle edits have to survive.
- **Debug APK only, arm64 only, today.** A release APK must be signed, and the
keystore has to be generated by the operator and never pass through CI logs or
an agent session — the same constraint recorded for the updater key (task 2136).
Gradle's throwaway debug keystore needs nothing from anyone, so this lane proves
the app compiles and packages. arm64 is every real device; the image carries the
other three ABIs, so widening is a one-word change.
- **Green means it BUILT.** Like the Windows lane, a Linux runner cannot execute
the artifact. Nothing here proves the app runs, renders, or is usable by finger.
- **TLS is the likely first failure.** `reqwest` is pinned to `native-tls`, which
is deliberate for the Windows lane (it resolves to `schannel`, keeping C/asm out
of the cross-compile). On Android it resolves to **OpenSSL**, which must be
cross-compiled per ABI — hence `perl` + `make` in the image. If that proves
painful, the fix is a target-specific dependency block selecting `rustls` for
Android only, leaving the Windows lane's reasoning untouched.
## Formatting the Rust lane before pushing
`cargo fmt --check` runs in CI and had failed on four consecutive desktop pushes