PowerShell: `New-Object` `-ArgumentList` vs calling constructor vs type casting

user_8457586 0
Viewed 395

I'm just starting to learn some basics of PowerShell and can't get my head around New-Object and type casting. For example this:

# all of these yield the same
New-Object System.Net.Sockets.TcpListener -ArgumentList 5000
New-Object System.Net.Sockets.TcpListener(5000)
[System.Net.Sockets.TcpListener]500  # this works

# all of these yield the same
New-Object -TypeName System.Net.Sockets.TCPClient -ArgumentList "8.8.8.8", 53
New-Object System.Net.Sockets.TCPClient("8.8.8.8", 53)
[System.Net.Sockets.TCPClient]::New("8.8.8.8", 53)
# but this doesn't work
# [System.Net.Sockets.TCPClient]("8.8.8.8", 53) # does not work
  • What is the difference between using -ArgumentList and using (arg1,...)? Is it that the name of the type and the (arg1,...) are interpreted as two different arguments to New-Object even when there is no space between them, and so the (arg1,...) is assigned as the value of -ArgumentList?
  • Why doesn't casting work for the second example of System.Net.Sockets.TCPClient? The only difference I see is that it takes multiple arguments, whereas System.Net.Sockets.TcpListener takes a single argument. But why can PowerShell cast an integer to a System.Net.Sockets.TcpListener but not an array to a System.Net.Sockets.TCPClient by calling its constructor? I.e. why aren't these two equivalent:
    • [System.Net.Sockets.TCPClient]::New("8.8.8.8", 53) # works
    • [System.Net.Sockets.TCPClient]("8.8.8.8", 53) # does not work
powershell
1 Answers

What is the difference between using -ArgumentList and using (arg1, ...)?

  • Do not use (arg1,...) it is pseudo method syntax that only happens to work - see the bottom section of this answer.

  • Instead, use arg1, ... - no parentheses, whitespace before the list; that is the positionally implied equivalent of -ArgumentList arg1, ... (or -Args arg1, ...)

Why doesn't casting work for the second example of System.Net.Sockets.TCPClient?

Casting only works in two variants:

  • With a scalar that matches a single-argument constructor or, in the case of a string, if the target type has a static .Parse() method.

  • With a hash table (@{ ... }) whose entries' keys match properties of the target type and assuming that the target type has a parameterless public constructor.

Therefore you cannot cast from an array of values.

See this answer for more information.

user_453750
Related