main
3
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
  shadowdaoandClaude Sonnet 5
|
b5c91c72e5 |
macOS: real .plugin bundle packaging, libobs @rpath fixup, hard-fail CI check
Closes the "macOS packaging gap" documented in README.md: CI built a real
adapter module on macOS but staged it as a bare streamer-tools-camera.so,
which OBS.app cannot load (it needs a <name>.plugin bundle), and otool -L
showed the libobs dependency as the relative path
libobs/libobs.framework/Versions/A/libobs instead of an @rpath reference.
cmake/macos/helpers.cmake adapts (not vendors as-is) obs-plugintemplate's
cmake/macos/helpers.cmake: upstream builds the bundle almost entirely
through XCODE_ATTRIBUTE_* properties and its own CI drives that with
`xcodebuild -project <name>.xcodeproj`, but this project's home-mac runner
has only the Command Line Tools, not Xcode.app (already established by
buildspec_common.cmake's CI-iteration-1 comment), and the whole project
builds with Ninja end to end. So this reimplements the same outcome --
Contents/MacOS, Contents/Resources, a real Info.plist -- with CMake's own
generator-agnostic BUNDLE/BUNDLE_EXTENSION/MACOSX_BUNDLE_INFO_PLIST/
MACOSX_PACKAGE_LOCATION target properties (verified working under Ninja via
a dry-run configure with APPLE spoofed), and does by hand what upstream gets
from Xcode's embed/codesign build phases:
- copies obs-adapter/data/** into Contents/Resources/**, since OBS's
AddExtraModulePaths() (UI/window-basic-main.cpp) passes
Contents/Resources as a macOS module's *data path* -- not a sibling
data/ the way Linux/Windows work -- so the locale ini has to land at
Contents/Resources/locale/en-US.ini for OBS_MODULE_USE_DEFAULT_LOCALE
to find it
- copies the LiveKit runtime dylibs into Contents/Frameworks
- fixes up the libobs dependency: cmake/macos/fixup-libobs-rpath.sh
rewrites the relative install name the from-source libobs build records
to @rpath/libobs.framework/Versions/A/libobs (LiveKit's own dylibs
already record @rpath references, confirmed in prior CI otool -L
output, so only libobs needs the rewrite)
- gives the plugin binary two LC_RPATH entries via INSTALL_RPATH:
@loader_path/../Frameworks (this bundle's own Frameworks, for LiveKit)
and @executable_path/../Frameworks (OBS.app/Contents/Frameworks, for
libobs.framework -- @executable_path is always relative to the host
process's main executable, not this dlopen'd bundle)
obs-adapter/CMakeLists.txt: calls stplugin_macos_finalize_bundle() before
staging, and corrects the staged layout for macOS -- OBS's module search
wants the whole <name>.plugin dropped directly into .../obs-studio/plugins/,
not nested under a bin/ subdirectory the way Linux/Windows are, so
build/package/ now holds the bundle at its own top level on macOS instead of
build/package/bin/<name>.plugin.
.gitea/scripts/macos-build.sh (this project's macOS steps were factored out
of build.yml into this shared script, used by both build.yml and
release.yml, in a concurrent commit -- rebased onto that): the "Show what
was built" section now hard-fails unless a real .plugin bundle (with
Info.plist, bundled LiveKit dylibs, and locale data) was produced, the
libobs dependency resolves via @rpath rather than the old relative path,
and both LC_RPATH entries are present -- closing the same
false-positive-green gap this project's Windows find_package(libobs)
incident just exposed, which this check previously did not cover (it only
checked for a bare .so). release.yml's macOS packaging step already globs
for any `*.plugin` under build/ (added in the concurrent commit,
anticipating this fix), so it picks up the new layout with no change
needed.
Verified: Linux configure/build/ctest (STPLUGIN_BOOTSTRAP_OBS=OFF) still
passes 6/6, unaffected -- the new macOS CMake logic is fully guarded by
if(APPLE). The bundle/Info.plist/Resources-mapping logic itself was dry-run
verified by configuring a throwaway CMake project with APPLE spoofed to
TRUE, confirming buildspec.json values substitute correctly into Info.plist
and obs-adapter/data/locale/en-US.ini maps to Resources/locale/en-US.ini.
Not verified: an actual macOS build/link/otool pass, or loading the result
in real OBS.app -- this is a Linux sandbox with no way to do either: this
commit is going to the home-mac CI runner to get that verification next.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RL8abRmgFXkVASHkkqiJbE
|
||
|
shadowdaoandClaude Sonnet 5
|
f281b4c382 |
ci: keep the synthesised macOS SDK symlink out of the dependency directory
libobs now builds AND installs cleanly on macOS -- libobs.framework, its
headers, and libobsConfig/libobsTargets all land. The configure then fell over
one step later, in the template's own quarantine-clearing step:
xattr: [Errno 13] Permission denied:
'.../.deps/sdk/MacOSX.platform/Developer/SDKs/MacOSX26.5.sdk'
`xattr -r -d com.apple.quarantine "${dependencies_dir}"` recurses, so it
followed the SDK symlink into the read-only system SDK. The symlink moves to
the build directory, where that sweep never reaches it. The xattr call itself
also stops being fatal: clearing quarantine on downloaded dependencies is a
convenience, and it should not be able to take the whole configure down.
Separately, in the adapter: the last-frame-geometry marker is cleared whenever
the session leaves Connected, so the next stream logs its first frame again.
Verified in the headless libobs harness by switching the selected camera to a
dark slot and back -- previously the switch back was silent because the
resolution had not changed, so the log stopped answering "did video come
back". It now reads:
connected to ws://... watching cam-test
video frame 640x360 I420
--- switching camera to 'other-cam'
connected to ws://... watching other-cam
--- switching camera back to 'cam-test'
connected to ws://... watching cam-test
video frame 640x360 I420
status after switch-back: connected
which also confirms the whole change-settings path: each switch mints a fresh
obs:<room>:<nonce> identity and reconnects, with no crash and no stale frame.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RL8abRmgFXkVASHkkqiJbE
|
||
|
shadowdaoandClaude Sonnet 5
|
8494485351 |
ci: build the real OBS adapter on all three platforms
Two things: unbreak the Windows compile, and give macOS/Windows a libobs.
MSVC fix. test_json.cpp failed to compile on the Windows runner with
"a universal-character-name specifies an invalid character" and "illegal
escape sequence" -- MSVC still forms escape sequences and
universal-character-names INSIDE raw string literals, which it must not.
Every JSON input containing a backslash is now built by string concatenation
from a single kBS constant, which also sidesteps the separate murky corner of
translation phase 1 where a doubled backslash immediately followed by 'u' has
historically been treated inconsistently. Same 158 checks, no behaviour
change.
OBS SDK bootstrap for macOS/Windows. Adopts obsproject/obs-plugintemplate's
buildspec machinery -- buildspec.json plus cmake/common/buildspec_common.cmake
and cmake/{macos,windows}/buildspec.cmake -- so those two platforms get a real
libobs and build the actual plugin module instead of only the core library.
Linux is untouched and still uses Ubuntu's libobs-dev
(-DSTPLUGIN_BOOTSTRAP_OBS=OFF); the bootstrap only runs where there is no
system package.
Trimmed against upstream, each change recorded in the file that makes it:
- qt6 is dropped from dependencies_list on both platforms. The properties UI
is plain obs_properties_* and nothing here links Qt.
- The OBS sub-build builds and installs the `libobs` target, not
`obs-frontend-api`. Building the frontend API is what would drag Qt back in,
and this plugin never calls it.
- The sub-build is configured with ENABLE_UI=OFF and ENABLE_SCRIPTING=OFF as
well as upstream's ENABLE_FRONTEND=OFF: the pinned OBS predates
ENABLE_FRONTEND and gates its Qt-dependent UI on ENABLE_UI, so without this
it configures the whole OBS UI and demands Qt anyway.
- Only the Release configuration is built and installed, not Debug as well.
Nothing consumes a debug libobs and it doubles the slowest CI step.
- Only the dependency-acquisition modules are vendored. The template's
compilerconfig/defaults/helpers/xcode modules drive its own target and
bundle layout, which this project does not use.
obs-studio is pinned to 30.0.2, deliberately low: OBS refuses to load a module
built against a NEWER libobs than the one running it and accepts older ones, so
this pin IS the minimum OBS version users need. 30.0.2 is also exactly what
Ubuntu 24.04's libobs-dev ships, which puts all three platforms on one floor,
and it supports the modern CMake layout the bootstrap drives via
-DOBS_CMAKE_VERSION=3.0.0. prebuilt is obs-deps 2023-11-03 with the hashes
obs-studio 30.0.2's own buildspec.json publishes; the obs-studio source
archive hashes were computed from the GitHub tag archives.
The workflow also prints what was actually produced on each platform (ldd /
otool / dir over build/package) and uploads it as an artifact, so "does this
even link against libobs" is answered by CI output rather than assumed.
Verified locally: the Linux path is unchanged by all of this -- a fresh
configure still finds libobs-dev, and ctest is 6/6. The macOS and Windows
bootstrap can only be verified by CI; that is what this push is for.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01RL8abRmgFXkVASHkkqiJbE
|