lp:~raof/mir/client-release-simplifications
- Get this branch:
- bzr branch lp:~raof/mir/client-release-simplifications
Branch merges
- Cemil Azizoglu (community): Needs Fixing
- Andreas Pokorny (community): Approve
- Mir CI Bot: Needs Fixing (continuous-integration)
-
Diff: 352 lines (+122/-59)8 files modifiedinclude/test/mir/test/validity_matchers.h (+8/-0)
src/client/atomic_callback.h (+15/-10)
src/client/mir_buffer_stream_api.cpp (+8/-1)
src/client/mir_connection.cpp (+11/-46)
src/client/mir_connection.h (+0/-2)
tests/acceptance-tests/staging/test_render_surface.cpp (+45/-0)
tests/mir_test/CMakeLists.txt (+4/-0)
tests/mir_test/validity_matchers.cpp (+31/-0)
Branch information
Recent revisions
- 3878. By Chris Halse Rogers
-
Eagerly destroy MirPresentation
Chain resources in _release() Just like MirBufferStream and MirRenderSurface, there's no need to wait for the
server to acknowledge. - 3877. By Chris Halse Rogers
-
Eagerly destroy MirBufferStream resources in _release()
Like MirRenderSurface, there's nothing to do client-side after sending the
release RPC message. Simplify things by deleting up-front. - 3876. By Chris Halse Rogers
-
Fix unsafe locking in AtomicCallback.
AtomicCallback was calling the underlying callback while holding a lock. Someone
clearly knew this was a bad idea, because they changed it to a std::recursive_mutex,
but that's not the only way this could deadlock.Since I've just seen a deadlock caused by this, fix it :)
- 3875. By Chris Halse Rogers
-
Eagerly destroy RenderSurface resources in _release()
There's no need to wait for the server to acknowledge our RPC request.
The client-side never uses any of the RenderSurface resources after sending
the release request, and there's no meaningfull error handling possible. - 3873. By Daniel van Vugt
-
PosixTimestamp: Add missing operators >= <=.
Approved by Alan Griffiths, mir-ci-bot.
Branch metadata
- Branch format:
- Branch format 7
- Repository format:
- Bazaar repository format 2a (needs bzr 1.16 or later)
- Stacked on:
- lp:mir