Closed strk closed 4 years ago
@SPlanzer it looks like the monkey-patch solution works, waiting for https://github.com/qgis/QGIS/issues/36187 to be fixed upstream...
Great CI is passing without the second test rest_runner but this does not see the python unit tests reported in the github action terminal
e.g.
Run make docker-qgis-test
docker-compose up -d
gazetteer_db_1 is up-to-date
gazetteer_qgis_1 is up-to-date
docker-compose exec -T qgis sh -c "/usr/bin/xvfb-run -- /scripts/testrunner.sh tests_directory.run_tests.run_test_modules"
Running test tests_directory.run_tests.run_test_modules ...
Warning: QStandardPaths: XDG_RUNTIME_DIR not set, defaulting to '/tmp/runtime-root'
QGIS Test Runner Inside - starting the tests ...
QGIS Test Runner - Trying to import tests_directory.run_tests.run_test_modules
QGIS Test Runner - Trying to import tests_directory.run_tests
QGIS Test Runner Inside - executing function <function run_test_modules at 0x7f0b0337f7b8>
test_metadata (test_metadata.TestMetadata) ... ok
test_click_search (test_ui.TestUi) ... ok
test_click_tab_text (test_ui.TestUi) ... ok
test_editordockwidget_is_docked (test_ui.TestUi) ... ok
test_layers_load_on_start (test_ui.TestUi) ... ok
test_tools_disabled_on_start (test_ui.TestUi) ... ok
test_tools_enabled_on_start (test_ui.TestUi) ... ok
----------------------------------------------------------------------
Ran 7 tests in 1.153s
OK
Finished running test tests_directory.run_tests.run_test_modules (codes: IS_DEAD=1 IS_FAILED=1 IS_PASSED=0).
Thanks for testing @SPlanzer, I've now force-pushed an amended commit fixing that problem
@strk Does this work nicely?
Now that the PR in QGIS is merged, do you still need this PR? Can't you use the new docker image generated?
@Gustry it has to be checked, if new QGIS images have been updated we should be fine.
I'm not an expert on TTY etc, but I tried and it didn't work. Can you let me know? I will investigate more soon.
We have to figure out which docker tag contains the upstream fix
I'm testing it, 3.10.7 should have the fix
Ok, 3.10.6 tag is still not available, see: https://hub.docker.com/r/qgis/qgis/tags
We'll have to wait for upstream to publish a new image
@strk https://github.com/qgis/QGIS/pull/36189 has been merged on master. So https://hub.docker.com/layers/qgis/qgis/latest/images/sha256-3f0ba39b9f130fde1b18b8b3e259f5faa8c3feb2b67d9d95d348ca875691e2cc?context=explore should be ok, no?
Yes, "latest" would be good, but "latest" is a moving target and we'd prefer (for stability) to have a stable one (a final tag). Also we want to be sure about the actual version we're using, I think.
We can, however, take the base image tag as a parameter, to experiment various tags, would you be willing to send a PR in that direction ?
Yes, "latest" would be good, but "latest" is a moving target and we'd prefer (for stability) to have a stable one (a final tag).
Yes I understand, it make sense to keep a known version of course.
We can, however, take the base image tag as a parameter, to experiment various tags, would you be willing to send a PR in that direction ?
I already gave it a try on my side, but it was unsuccessful for now. That's why I wanted to know if you had a look.
With doing a test matrix with a parameter, you could do a PR against your repo to try the qgis:latest image to check if your patch in upstream was working for you.
Yes, see #133 for a PR
@strk Have you got some progress by just trying in a PR, before 3.14 is out to try the latest tag?
@Gustry no, I did not test, but you can now do easily, as https://github.com/linz/gazetteer/pull/133 was merged. Can you give it a try while I'm on a low-bandwidth network ?
Can't stand the /dev/tty issue