Changes for page Test Speedup
Last modified by chrisby on 2025/03/08 11:39
Summary
-
Page properties (1 modified, 0 added, 0 removed)
Details
- Page properties
-
- Content
-
... ... @@ -20,10 +20,10 @@ 20 20 21 21 **Synchronous Testing** 22 22 23 -A simple TDD workflow is to write new code, runtests locally, wait for them to finish,and if they pass,move on. To avoid long wait times, yourunonly a few very fast tests.Thissfinewhenyou are working on isolated code andusing unit tests,but as soon as integrationofthenewcodewiththeoldcode comes into play,itbecomes a problem. Youhave two bad choices:either you run only a few fast tests and do not use the full power of your test suite, possibly missing bugs that would beeasier to fix if they werecaught earlier,or you run all thetestslocallyand are unproductive for along timewhile waiting for themto finish.This problem can be solved with asynchronous testing.23 +A simple TDD workflow is to write new code, execute tests locally, wait for them to finish and if they pass then going on. To avoid long waiting times being unproductive, you only execute a few, very fast tests sacrificing that all tests check your latest changes. This is tolerable when working on yet isolated code and working with unit tests but as soon as integration with the rest of the code comes into play this becomes a problem. Eit 24 24 25 25 **Asynchronous Testing** 26 26 27 - YoushouldhaveDevOps infrastructurewhichwhenpushing code to thecode repositorytriggersaCI pipelineexecuting all tests.Doing that enablesyouto directlygo on working withouttheneedto wait minutesfortheteststofinish. If the CI pipelinesucceeds, theomprehensivetest suite proofedyour codeto be fine. If the CI pipelinefails,you should get a notification like an SMS orEmail, abort yourcurrentwork immediately to fix the problem. Pushthe fixagainand continueworking withoutwaitingfor anyteststo finish.27 +is a workflow that works well when the test take a few seconds only. This has the disadvantage that you only check your code changes for correctness against just a few very fast tests. 28 28 29 -It is no tunusualthatmanyCI pipelinesrun at thesame time for the samedeveloper.Although thisposesaquite huge29 +Instead of waiting for your tests to finish locally, you should have a DevOps infrastructure which triggers a CI pipeline when pushing the code executing all tests. Doing that enables you to directly go on working without the need to wait minutes for the tests to finish. In case the CI pipeline fails, you should get a notification to fix the CI pipeline immediately. This enables quite comprehensive testing, even having the same testing jobs running in parallel, even long taking ones.