oak-team/oak-editor
3
1
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

refactor(cli,worker): cut liboakengine, link module rlibs directly (M14 R2)

Browse Source 
- oak-cli: new engine.rs assembly layer maps every facade call to
  module Rust APIs (oaknode graph/serializer, oaktimeline commands,
  oakrender ticket arena, oaktask ExportTask, oakcommon config); the
  ffi/optional/host layers and build.rs link config are gone
- oak-worker: the worker session + POSIX shm transport moved into the
  crate (oakrender backend + serde_json control plane); no dylib
- both binaries carry zero liboakengine references (otool verified);
  tests green (30 cli / 41 worker)
This commit is contained in:
Mike-Solar
2026-08-16 21:16:27 +08:00
parent 8a2e45225f
commit a45a7af2ac
24 changed files with 3713 additions and 3196 deletions
Download Patch File Download Diff File Expand all files Collapse all files
crates/oak-cli/tests/cli.rs
+13 -15
View File
@@ -18,10 +18,9 @@
//! `oak_cli_info`/`oak_cli_render`/`oak_cli_probe`/`oak_cli_transcode`
//! equivalents).
//!
//! The binary is a pure C-ABI consumer of `liboakengine` (see src/ffi.rs),
//! so these tests need the built dylib present: run `cargo build -p
//! oakengine` BEFORE `cargo test -p oak-cli` (the link happens at build
//! time, the load at runtime through the rpath build.rs installs).
//! The binary links the oak* module crates directly (M14 R2; see
//! src/engine.rs) — no `liboakengine` dylib is needed at build or run
//! time, so `cargo test -p oak-cli` stands alone.
//!
//! The data-producing paths run against the repo fixtures
//! (`tests/project_with_footage.ove`, `tests/demo.mp4` — real H.264/AAC
@@ -29,10 +28,10 @@
//! codes (0 success, 1 general error, 2 rendering unavailable, 64 usage
//! error), and the argument-validation paths assert the C++ messages.
//!
//! The engine's footage probe records the decoder id but drops the
//! codec's stream descriptions (oaknode module gap), so the probe output
//! carries real stream counts of 0 and a 0 duration until the module
//! fills them in — the assertions pin the real contract.
//! The module probe records the decoder id but drops the codec's stream
//! descriptions (oaknode module gap), so the probe output carries real
//! stream counts of 0 and a 0 duration until the module fills them in —
//! the assertions pin the real contract.
use std::path::{Path, PathBuf};
use std::process::Command;
@@ -78,9 +77,8 @@ fn run(args: &[&str]) -> (i32, String, String) {
)
}
/// stderr without dyld's objc class-duplication notices (the engine
/// dylib embeds FFmpeg's libavdevice, which collides with the host's on
/// macOS).
/// stderr, kept for the historical dyld objc notices (the module-linked
/// binary no longer embeds FFmpeg's libavdevice, so it is a pass-through).
fn real_errors(stderr: &str) -> String {
stderr
.lines()
@@ -119,7 +117,7 @@ fn help_prints_the_cpp_usage_text_and_exits_zero() {
!real_errors(&stderr).contains("error:"),
"stderr: {stderr}"
);
assert!(stdout.starts_with("oak-cli - headless consumer of the liboakengine C ABI\n"));
assert!(stdout.starts_with("oak-cli - headless consumer of the oak editor modules (direct Rust ABI)\n"));
assert!(stdout.contains("oak-cli transcode <input_media> <out> [width] [--format ppm|mp4]"));
assert!(stdout.contains("Exit codes:"));
assert!(stdout.contains("64 usage error"));
@@ -312,9 +310,9 @@ fn transcode_the_image_fixture_to_ppm_frames() {
}
#[test]
fn transcode_mp4_reports_the_unwrapped_exporter_family() {
// The facade wraps the exporter family (oakengine_export_render),
// so the mp4 path renders and writes a real file.
fn transcode_mp4_writes_a_real_mp4() {
// The module export task (crate::engine::export_sequence) drives the
// mp4 path end to end and writes a real file.
let dir = TempDir::new("transcode_mp4");
let image = fixture_image();
let out = dir.0.join("out.mp4");