-
Notifications
You must be signed in to change notification settings - Fork 12
Description
We still can't run the UI tests and this makes the whole CI run super slow (as each test runs into a timeout).
Locally the tests mostly work:
- debian testing: 84%: Checks: 113, Failures: 0, Errors: 18
- opensuse tumbleweed: ...
The test setup believes it has successfully setup the xvfb:
0:24:22.152876977 72611 0x55f86d997c70 INFO bt-check tests/bt-check-ui.c:221:check_setup_test_display: display 0x55f86d9e85e0,":0" is active
It is not even clear if the xvfb setup is to blame. A lot of the tests have these as the last log lines:
0:26:02.382873958 72703 0x55f86d997c70 INFO bt-edit src/ui/edit/main-page-sequence.c:1413:sequence_table_init: preparing sequence table
0:26:02.383991664 72703 0x55f86d997c70 INFO bt-core src/lib/core/settings.c:408:bt_settings_make: return cached settings object 0x55f86dd10560,ref_ct=3,floating=0
0:26:02.384494252 72703 0x55f86d997c70 INFO GST_PLUGIN_LOADING ../gst/gstplugin.c:987:_priv_gst_plugin_load_file_for_registry: plugin "/usr/lib/x86_64-linux-gnu/gstreamer-1.0/libgstplayback.so" loaded
0:26:02.384515702 72703 0x55f86d997c70 INFO GST_ELEMENT_FACTORY ../gst/gstelementfactory.c:489:gst_element_factory_create_with_properties: creating element "playbin"
The test itself seem to actually work and finish quickly (we see cleanup markers in the log). It feels more like the something in the test hardness hangs (e.g. taking down the xvfb server)
Completely alternative idea:
We could try to use a matrix to run groups of tests in parallel.
https://stackoverflow.com/questions/59175332/using-output-from-a-previous-job-in-a-new-one-in-a-github-action
https://docs.github.com/en/actions/using-jobs/using-a-matrix-for-your-jobs