[hoanle0126]

Proposed changes

• Remove existing pry_load_appium_txt and reload singleton methods on Pry before redefining them in Appium::Console.setup to avoid method redefined warnings.
• Replace any existing 'Release session hook' :after_session hook before registering it in Appium::Console.start, and guard x with respond_to?(:x, true) so the :after_session hook does not raise NameError if a session failed to start or promote driver methods.
• Check for both :newCommandTimeout and :'appium:newCommandTimeout' in lib/start.rb before injecting the default newCommandTimeout: 999_999 capability.

Types of changes

• Bugfix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)

[KazuCocoa]

[P2] Assert that the release hook completes without an error

Pry.hooks.exec_hook catches hook exceptions and stores them in Pry.hooks.errors, so this assertion still passes if the new respond_to?(:x, true) guard is removed: the closing message is printed before x raises NameError. I reproduced this behavior with Pry 0.14.2's hook implementation. Please invoke the registered hook directly with Pry.hooks.get_hook(:after_session, 'Release session hook').call(output, binding, Pry.new), or assert that executing it adds no hook errors, so the test detects the missing-driver regression.

[hoanle0126]

Updated test_after_session_hook_handles_missing_driver_and_repeated_start to invoke Pry.hooks.get_hook(:after_session, 'Release session hook').call(output, binding, Pry.new) directly and assert Pry.hooks.errors is empty. Thanks @KazuCocoa!