This change updates ModuleIntrinsics.GetModulePath to handle the case where the Windows PowerShell module path is already in the environment's PSModulePath or when launched from a different version of PowerShell. Previously, GetModulePath would append $PSHOME\Modules to the PSModulePath after removing the path for the launching version without considering the Windows PowerShell module path. The result, was the Windows PowerShell modules were found first and loaded incompatible modules; such as the built-in modules. The change detects the Windows PowerShell module path and inserts $PSHOME\Modules path before it. The new test simulates launching from a different version of pwsh that has already added the Windows PowerShell module path. Fixes #7679 |
||
---|---|---|
.. | ||
engine | ||
Host | ||
Installer | ||
Language | ||
Modules | ||
Provider | ||
SDK | ||
README.md |
Pester Testing Test Guide
Also see the Writing Pester Tests document.
Running Pester Tests
Go to the top level of the PowerShell repository and run: Start-PSPester
inside a self-hosted copy of PowerShell.
You can use Start-PSPester -Tests SomeTestSuite*
to limit the tests run.
Testing new powershell
processes
Any launch of a new powershell
process must include -noprofile
so that
modified user and system profiles do not causes tests to fail. You also must
take care to call the development copy of PowerShell, which is not the first
one on the path.
Example:
$powershell = Join-Path -Path $PsHome -ChildPath "pwsh"
& $powershell -noprofile -command "ExampleCommand" | Should Be "ExampleOutput"
Portability
Some tests simply must be tied to certain platforms. Use Pester's
-Skip
directive on an It
statement to do this. For instance to run
the test only on Windows:
It "Should do something on Windows" -Skip:($IsLinux -Or $IsMacOS) { ... }
Or only on Linux and OS X:
It "Should do something on Linux" -Skip:$IsWindows { ... }
Pending
When writing a test that should pass, but does not, please do not skip or delete
the test, but use It "Should Pass" -Pending
to mark the test as pending, and
file an issue on GitHub.