Cannot easily generate bash RunTests scripts from Windows, and vice versa #69555
Comments
ghost
added
the
|
Tagging subscribers to this area: @dotnet/area-infrastructure-libraries Issue DetailsDescriptionThere are cases where we might want to do a build on a Linux/Mac system, but the Helix queue for execution is Windows (or vice versa). In fact, this is already the case for Android, where our device queue is on Windows but are builds are all on Linux, or Wasm where our builds are on Linux but the browser queue is Windows. However, for these cases, we rely on the Helix SDK to generate run scripts (e.g. submitting a list of APK files to Helix and allowing Helix to decide what the test runner scripts should be). There may be cases where we need to generate a script, as we do for most test runner cases, but build/submit a script which doesn't match the build OS. Rudimentary support for this is included in #66147 but relies on a yaml parameter to switch behaviour. IMHO we should probably always generate both .sh and .cmd test scripts for all cases where that could make sense (e.g. Android) Reproduction StepsBuild tests subset for Android, look in a test directory Expected behaviorTest runner script for both Windows and Mac/Linux, so the test can be easily run on a second machine Actual behaviorOnly Runtests.sh exists Regression?No response Known WorkaroundsNo response ConfigurationNo response Other informationNo response
|
|
@directhex I wonder - some of the coreclr tests have native components that need to be built on Unix. How would you handle that aspect if you wanted to build the managed stuff and generate the scripts on Windows? |
|
Good question. Not sure! |
carlossanlop
removed
the
untriaged
Description
There are cases where we might want to do a build on a Linux/Mac system, but the Helix queue for execution is Windows (or vice versa). In fact, this is already the case for Android, where our device queue is on Windows but are builds are all on Linux, or Wasm where our builds are on Linux but the browser queue is Windows. However, for these cases, we rely on the Helix SDK to generate run scripts (e.g. submitting a list of APK files to Helix and allowing Helix to decide what the test runner scripts should be).
There may be cases where we need to generate a script, as we do for most test runner cases, but build/submit a script which doesn't match the build OS. Rudimentary support for this is included in #66147 but relies on a yaml parameter to switch behaviour.
IMHO we should probably always generate both .sh and .cmd test scripts for all cases where that could make sense (e.g. Android)
Reproduction Steps
Build tests subset for Android, look in a test directory
Expected behavior
Test runner script for both Windows and Mac/Linux, so the test can be easily run on a second machine
Actual behavior
Only Runtests.sh exists
Regression?
No response
Known Workarounds
No response
Configuration
No response
Other information
No response
The text was updated successfully, but these errors were encountered: