potiuk commented on a change in pull request #8383: Remove steps from travis
URL: https://github.com/apache/airflow/pull/8383#discussion_r409041753
 
 

 ##########
 File path: .github/workflows/static-tests.yml
 ##########
 @@ -0,0 +1,36 @@
+name: "static tests"
+on: [pull_request, push]
+env:
+  TRAVIS_BRANCH: "master"
+jobs:
+  statics-tests:
+    name: Static tests
+    runs-on: ubuntu-latest
+    env:
+      PYTHON_VERSION: 3.6
+      AIRFLOW_MOUNT_SOURCE_DIR_FOR_STATIC_CHECKS: "true"
+      TRAVIS_JOB_NAME: "Static"
+    strategy:
+      matrix:
+        scripts: [ci_run_static_checks_pylint_tests.sh,
+                  ci_mypy.sh,
+                  ci_fix_ownership.sh,
 
 Review comment:
   Been there done that @dimberman .
   
   I think this will not increase the speed. It might even slow it down. 
   
   There is quite an overhead introduced by creating a new runner (it works 
similarly in Travis and GA last time I checked). Vast majority of the tests - 
including mypy, llake8, bat_tests are fast and it makes no sense to spin 
separate worker just to run the single static check.. The ones that take vast 
(like 90%) of time are pylint_main and pylint_test), I experimented with it 
before and the best overall execution time was what was there before:
   
   1, job - run all tests but pylint_tests
   2, Run pylint_tests
   
   Further fragmentation introduces much more overhead for worker creation and 
elapsed time to execute the builds was much longer. With this approach you can 
gain at most single seconds in the best case. 
   
   What would ultimately help if is we finish all pylint tests and introduce 
paralllelism in pylint tests (the tests can run muich faster but currently we 
have uncertainty of cycles introduced because some of our code is excluded from 
pylint and contains cycles, Those cycles are randomly appearing when people 
make their PRs so Pylint tests are now set to run without parallelism.
   
   
    

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
[email protected]


With regards,
Apache Git Services

Reply via email to