When handling file redirection for CommandExpression, we don't call 'DoComplete' on the underlying PipelineProcessor of the FileRedirection object, and thus the EndProcessing method is not called on Out-File, which causes different behaviors between <expr> > out.txt and <expr> | Out-File out.txt.
The fix is to make sure 'DoComplete' is called after the stream output has been written to the redirection pipe.
Also fix another issue
This PR also fixes an issue that could mess up restoring the original pipes. Here is the repro:
PS> 1 *> b.txt > a.txt; 123
Cannot perform operation because object "PipelineProcessor" has already been disposed
The root cause is that we don't always restore pipes in the correct order. Please see the code changes in Compiler.cs for more details.
Fix#4812
This round of cleanup focuses on the following areas
- Tab completion code
- InitialSessionState
- Compiler/MiscOps
- Utils/SessionStateFunctionAPIs
Changes in the rest affected files are mainly some corresponding changes due to the cleanup work in the above areas.
When dotting a script cmdlet, the locals of automatic variables from the `PSScriptCmdlet` is not set up in the current scope before parameter binding. The fix is to push the locals in `CommandProcessor.OnSetCurrentScope` and pop them in `CommandProcessor.OnRestorePreviousScope`, which will be called from `SetCurrentScopeToExecutionScope` and `RestorePreviousScope` respectively.
Summary of changes:
1. When a new local scope is used, currently we set the locals for `CommandProcessor` right before parameter binding (in `BindCommandLineParametersNoValidation`); we set the locals for `ScriptCommandProcessor` in `Prepare`. I moved both to the constructor, right after the new scope is created so that the code is more consistent.
2. In `CmdletParameterBinderController.cs`, we set up the `PSBoundParameters` and `MyInvocation` variables in `HandleCommandLineDynamicParameters` again, which I think is unnecessary because this method is only called from `BindCommandLineParametersNoValidation`, where the setup is done for the first time.
3. Currently, the locals will be set for dotted script cmdlet in `EnterScope()` and `ExitScope()`. Now, that logic is moved to `OnSetCurrentScope` and `OnRestorePreviousScope` of `CommandProcessor`. This not only makes sure that locals are set before parameter binding, but also is consistent with the `ScriptCommandProcessor`.
GetMember/SetMember/InvokeMember operations on a COM object will generate code to unwrap the COM object if it's wrapped in PSObject. Due to a loose restriction, if you have a string wrapped to a PSObject with an ETS member of the same name, then the string PSObject will be unwrapped too and the ETS member will be lost. The fix is to use a more restricted rule by checking if the base object is a COM object.
- Fix PSScriptAnalyzer warnings of type PSAvoidUsingCmdletAliases for 'ForEach-Object' (alias is '%' or 'foreach')
- Fix PSScriptAnalyzer warnings of type PSAvoidUsingCmdletAliases for 'Where-Object' (alias is '?' or 'where')
- Fix PSScriptAnalyzer warnings of type PSAvoidUsingCmdletAliases for 'Select-Object' (alias is 'select')
- Fix PSScriptAnalyzer warnings of type PSPossibleIncorrectComparisonWithNull. Essentially, $null has to be on the left-hand side when using it for comparison.
- A Test in ParameterBinding.Tests.ps1 needed adapting as this test used to rely on the wrong null comparison
- Replace a subset of tests of kind '($object -eq $null) | Should Be $true' with '$object | Should Be $null'
Summary
----------
When the same script file gets executed in multiple Runspaces, the RuntimeType generated from the powershell class defined in the file will be shared among those Runspaces. For different Runspaces, different SessionState should be used to run the static methods.
Fix
---
We use the `SessionStateKeeper` to query for the correct `SessionState` to use for a static method call.
We already use `SessionStateKeeper` to track the `Runspace/SessionState` where a PowerShell class is defined, and thus we can leverage this for the static method invocation.
Implements support for backgrounding pipelines with &. Putting & at the end of a pipeline will cause the pipeline
to be run as a PowerShell job. When a pipeline is backgrounded a job object is returned. Once the pipeline is
running as a job, all of the normal job cmdlets can be used to manage the job. Variables (ignoring process-specific
variables) used in the pipeline are automatically copied to the job so
copy $foo $bar &
just works. The job is also run in the current directory instead of the user's home directory as is the case with Start-Job.Implement
This change moves powershell to .NET Core 2.0. Major changes are:
1. PowerShell assemblies are now targeting `netcoreapp2.0`. We are using `microsoft.netcore.app-2.0.0-preview1-001913-00`, which is from dotnet-core build 4/4/17. We cannot target `netstandard2.0` because the packages `System.Reflection.Emit` and `System.Reflection.Emit.Lightweight`, which are needed for powershell class, cannot be referenced when targeting `netstandard2.0`.
2. Refactor code to remove most CLR stub types and extension types.
3. Update build scripts to enable CI builds. The `-cache` section is specified to depend on `appveyor.yml`, so the cache will be invalidated if `appveyor.yml` is changed.
4. Ship `netcoreapp` reference assemblies with powershell to fix the issues in `Add-Type` (#2764). By default `Add-Type` will reference all those reference assemblies when compiling C# code. If `-ReferenceAssembly` is specified, then we search reference assemblies first, then the framework runtime assemblies, and lastly the loaded assemblies (possibly a third-party one that was already loaded).
5. `dotnet publish` generates executable on Unix platforms, but doesn't set "x" permission and thus it cannot execute. Currently, the "x" permission is set in the build script, `dotnet/cli` issue [#6286](https://github.com/dotnet/cli/issues/6286) is tracking this.
6. Replace the use of some APIs with the ones that take `SecureString`.
7. osx.10.12 is required to update to `netcoreapp2.0` because `dotnet-cli` 2.0.0-preview only works on osx.10.12.
8. Add dependency to `System.ValueTuple` to work around a ambiguous type identity issue in coreclr. The issue is tracked by `dotnet/corefx` [#17797](https://github.com/dotnet/corefx/issues/17797). When moving to newer version of `netcoreapp2.0`, we need to verify if this dependency is still needed.
* Refactor 'FindMatchingHandler' to get rid of recursion
Search state can be saved and used throughout the searching process without the recursion.
* Rank the exception types based on how specific they are
* Refactor code to find handler based on the rank of given handlers
When a PS class is defined in a module and the module gets reloaded, the class would still use the SessionState from the old module for execution, and thus it doesn't reflect changes to the module state during the reload. This fix is to make sure we always use the current EngineSessionState for PS class execution.
Our private implementation of FailFast was likely introduced
when each individual application needed it's own code to
get Watson reports. The CLR takes care of this for us now,
so we don't need our implementation.
Our method CheckForSevereException was also introduced in
the early days of the CLR - the exceptions it was checking
for aren't actually raised the CLR anymore, they just FailFast.
I removed them as there is a tiny bit of overhead (in code size)
and also in the generated code, e.g. dynamic sites called the method.
People use Out-Null despite much faster alternatives:
$null = Do-Stuff
[void] = Do-Stuff
Do-Stuff > $null
This change makes Out-Null work in roughly the same say as the above.
The optimization is to detect that we're calling the built-in Out-Null
cmdlet when invoking a pipeline from script (the change won't have any
effect in the PowerShell api). If we detect Out-Null, we rewrite the
pipe to look similar to `Do-Stuff > $null`.
There was a NullReferenceException after turning on constrained language mode
if creating a binding on a thread with no runspace.
The fix is to assume if there is no runspace that we are not in constrained language
mode, and also generate a binding restriction that assumes the same.
We use a lot of reflection, and when you rename a member (e.g. when running
the code formatter), it doesn't detect references where you're using reflection.
By using nameof(), we can now find these references (e.g. a build break) instead
of failing at runtime.
I (Jason Shirk) ran https://github.com/dotnet/codeformatter with the default rules, basically:
codeformatter /nocopyright "/c:DEBUG,UNIX,CORECLR" @files.rsp
This caused a few problems building, which were fixed up manually.
Notable changes:
`this.` is removed unless needed to disambiguate.
private instance fields are renamed to have a `_` prefix.
private static fields are renamed to have a `s_` prefix.
I left some projects alone (like PackageManagement) and also left some generated code alone.
Dongbo Wang