| Summary: | [GTK] run-gtk-tests should use the driver environment for checking the accessibility bus | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Product: | WebKit | Reporter: | Carlos Alberto Lopez Perez <clopez> | ||||||
| Component: | Tools / Tests | Assignee: | Carlos Alberto Lopez Perez <clopez> | ||||||
| Status: | RESOLVED FIXED | ||||||||
| Severity: | Normal | CC: | bugs-noreply, cgarcia, darin, lforschler | ||||||
| Priority: | P2 | ||||||||
| Version: | WebKit Nightly Build | ||||||||
| Hardware: | Unspecified | ||||||||
| OS: | Unspecified | ||||||||
| Bug Depends on: | |||||||||
| Bug Blocks: | 160736 | ||||||||
| Attachments: |
|
||||||||
Created attachment 286869 [details]
Patch
Created attachment 286871 [details]
Patch
trivial: fix identation
Comment on attachment 286871 [details] Patch View in context: https://bugs.webkit.org/attachment.cgi?id=286871&action=review > Tools/Scripts/run-gtk-tests:155 > + def _wait_for_accessibility_bus(self, timeout=5, env={}): Why are params optional if you are passing them always? If the timeout can't be configured, we don't need to pass it to the function, use 5 directly in GLib.timeout_add_seconds > Tools/Scripts/run-gtk-tests:191 > + if not self._wait_for_accessibility_bus(timeout=5, env=self._test_env): You don't need to use the name=value, just pass the values. But in this case, you are passing a hardcoded value that could be used directly by the function, and a member variable to a member function, so the function could simply use self._test_env Committed r205017: <http://trac.webkit.org/changeset/205017> |
Currently run-gtk-tests waits for the accessibility bus on D-Bus to appear like this: # We need to wait until the SPI bus is launched before trying to start the SPI # registry, so we spin a main loop until the bus name appears on DBus. loop = GLib.MainLoop() Gio.bus_watch_name(Gio.BusType.SESSION, 'org.a11y.Bus', Gio.BusNameWatcherFlags.NONE, lambda *args: loop.quit(), None) loop.run() The problems are: 1) It runs the check using the environment from the user rather than the environ that should be used for the tests. So the environment variables from the driver (after bug 160736) are not correctly set (like DISPLAY or WAYLAND_DISPLAY). This seems to be somehow relevant, because if the system don't has the environment variable DBUS_SESSION_BUS_ADDRESS defined, dbus is able to do the right thing if at least DISPLAY is set to the correct value. Otherwise it fails. 2) It don't has a timeout. In case the accessibility bus don't appears, it hangs forever.