Difference between PSObject, Hashtable, and PSCustomObject - powershell

Question

Can anybody explain the details? If I create an object using
$var = [PSObject]#{a=1;b=2;c=3}
and then I look for its type using getType() PowerShell tells me it's of type Hashtable.
When using Get-Member (alias gm) to inspect the object it's obvious that a hashtable has been created, since it has a keys and a values property. So what's the difference to a "normal" hashtable?
Also, what's the advantage of using a PSCustomObject? When creating one using something like this
$var = [PSCustomObject]#{a=1;b=2;c=3}
the only visible difference to me is the different datatype of PSCustomObject. Also instead of keys and value properties, a inspection with gm shows that now every key has been added as a NoteProperty object.
But what advantages do I have? I'm able to access my values by using its keys, just like in the hashtable. I can store more than simple key-value pairs (key-object pairs for example) in the PSCustomObject, JUST as in the hashtable. So what's the advantage? Are there any important differences?

Answer 1

One scenario where [PSCustomObject] is used instead of HashTable is when you need a collection of them. The following is to illustrate the difference in how they are handled:
$Hash = 1..10 | %{ #{Name="Object $_" ; Index=$_ ; Squared = $_*$_} }
$Custom = 1..10 | %{[PSCustomObject] #{Name="Object $_" ; Index=$_ ; Squared = $_*$_} }
$Hash | Format-Table -AutoSize
$Custom | Format-Table -AutoSize
$Hash | Export-Csv .\Hash.csv -NoTypeInformation
$Custom | Export-Csv .\CustomObject.csv -NoTypeInformation
Format-Table will result in the following for $Hash:
Name Value
---- -----
Name Object 1
Squared 1
Index 1
Name Object 2
Squared 4
Index 2
Name Object 3
Squared 9
...
And the following for $CustomObject:
Name Index Squared
---- ----- -------
Object 1 1 1
Object 2 2 4
Object 3 3 9
Object 4 4 16
Object 5 5 25
...
The same thing happens with Export-Csv, thus the reason to use [PSCustomObject] instead of just plain HashTable.

Answer 2

Say I want to create a folder. If I use a PSObject you can tell it is wrong by
looking at it
PS > [PSObject] #{Path='foo'; Type='directory'}
Name Value
---- -----
Path foo
Type directory
However the PSCustomObject looks correct
PS > [PSCustomObject] #{Path='foo'; Type='directory'}
Path Type
---- ----
foo directory
I can then pipe the object
[PSCustomObject] #{Path='foo'; Type='directory'} | New-Item

Answer 3

From the PSObject documentation:
Wraps an object providing alternate views of the available members and ways to extend them. Members can be methods, properties, parameterized properties, etc.
In other words, a PSObject is an object that you can add methods and properties to after you've created it.
From the "About Hash Tables" documentation:
A hash table, also known as a dictionary or associative array, is a compact data structure that stores one or more key/value pairs.
...
Hash tables are frequently used because they are very efficient for finding and retrieving data.
You can use a PSObject like a Hashtable because PowerShell allows you to add properties to PSObjects, but you shouldn't do this because you'll lose access to Hashtable specific functionality, such as the Keys and Values properties. Also, there may be performance costs and additional memory usage.
The PowerShell documentation has the following information about PSCustomObject:
Serves as a placeholder BaseObject when PSObject's constructor with no parameters is used.
This was unclear to me, but a post on a PowerShell forum from the co-author of a number of PowerShell books seems more clear:
[PSCustomObject] is a type accelerator. It constructs a PSObject, but does so in a way that results in hash table keys becoming properties. PSCustomObject isn't an object type per se – it's a process shortcut. ... PSCustomObject is a placeholder that's used when PSObject is called with no constructor parameters.
Regarding your code, #{a=1;b=2;c=3} is a Hashtable. [PSObject]#{a=1;b=2;c=3} doesn't convert the Hashtable to a PSObject or generate an error. The object remains a Hashtable. However, [PSCustomObject]#{a=1;b=2;c=3} converts the Hashtable into a PSObject. I wasn't able to find documentation stating why this happens.
If you want to convert a Hashtable into an object in order to use its keys as property names you can use one of the following lines of code:
[PSCustomObject]#{a=1;b=2;c=3}
# OR
New-Object PSObject -Property #{a=1;b=2;c=3}
# NOTE: Both have the type PSCustomObject
If you want to convert a number of Hashtables into an object where their keys are property names you can use the following code:
#{name='a';num=1},#{name='b';num=2} |
% { [PSCustomObject]$_ }
# OR
#{name='a';num=1},#{name='b';num=2} |
% { New-Object PSObject -Property $_ }
<#
Outputs:
name num
---- ---
a 1
b 2
#>
Finding documentation regarding NoteProperty was difficult. In the Add-Member documentation, there isn't any -MemberType that makes sense for adding object properties other than NoteProperty. The Windows PowerShell Cookbook (3rd Edition) defined the Noteproperty Membertype as:
A property defined by the initial value you provide
Lee, H. (2013). Windows PowerShell Cookbook. O'Reilly Media, Inc. p. 895.

Answer 4

One advantage I think for PSObject is that you can create custom methods with it.
For example,
$o = New-Object PSObject -Property #{
"value"=9
}
Add-Member -MemberType ScriptMethod -Name "Sqrt" -Value {
echo "the square root of $($this.value) is $([Math]::Round([Math]::Sqrt($this.value),2))"
} -inputObject $o
$o.Sqrt()
You can use this to control the sorting order of the PSObject properties (see PSObject sorting)

Answer 5

I think the biggest difference you'll see is the performance. Have a look at this blog post:
Combining Objects Efficiently – Use a Hash Table to Index a Collection of Objects
The author ran the following code:
$numberofobjects = 1000
$objects = (0..$numberofobjects) |% {
New-Object psobject -Property #{'Name'="object$_";'Path'="Path$_"}
}
$lookupobjects = (0..$numberofobjects) | % {
New-Object psobject -Property #{'Path'="Path$_";'Share'="Share$_"}
}
$method1 = {
foreach ($object in $objects) {
$object | Add-Member NoteProperty -Name Share -Value ($lookupobjects | ?{$_.Path -eq $object.Path} | select -First 1 -ExpandProperty share)
}
}
Measure-Command $method1 | select totalseconds
$objects = (0..$numberofobjects) | % {
New-Object psobject -Property #{'Name'="object$_";'Path'="Path$_"}
}
$lookupobjects = (0..$numberofobjects) | % {
New-Object psobject -Property #{'Path'="Path$_";'Share'="Share$_"}
}
$method2 = {
$hash = #{}
foreach ($obj in $lookupobjects) {
$hash.($obj.Path) = $obj.share
}
foreach ($object in $objects) {
$object |Add-Member NoteProperty -Name Share -Value ($hash.($object.path)).share
}
}
Measure-Command $method2 | select totalseconds
Blog author's output:
TotalSeconds
------------
167.8825285
0.7459279
His comment regarding the code results is:
You can see the difference in speed when you put it all together. The object method takes 167 seconds on my computer while the hash table method will take under a second to build the hash table and then do the lookup.
Here are some of the other, more-subtle benefits:
Custom objects default display in PowerShell 3.0

Answer 6

We have a bunch of templates in our Windows-PKI and we needed a script, that has to work with all active templates. We do not need to dynamically add templates or remove them.
What for me works perfect (since it is also so "natural" to read) is the following:
$templates = #(
[PSCustomObject]#{Name = 'template1'; Oid = '1.1.1.1.1'}
[PSCustomObject]#{Name = 'template2'; Oid = '2.2.2.2.2'}
[PSCustomObject]#{Name = 'template3'; Oid = '3.3.3.3.3'}
[PSCustomObject]#{Name = 'template4'; Oid = '4.4.4.4.4'}
[PSCustomObject]#{Name = 'template5'; Oid = '5.5.5.5.5'}
)
foreach ($template in $templates)
{
Write-Output $template.Name $template.Oid
}

Answer 7

Type-1: $PSCustomObject = [PSCustomObject] #{a=1;b=2;c=3;d=4;e=5;f=6}
Type-2: $PsObject = New-Object -TypeName PSObject -Property #{a=1;b=2;c=3;d=4;e=5;f=6}
The only difference between Type-1 & Type-2
Type-1 Property are displayed in same order as we added
Type-1 enumerates the data faster
Type-1 will not work with systems running PSv2.0 or earlier
Both Type-1 & Type-2 are of type “System.Management.Automation.PSCustomObject”
Difference between HashTable and PSCustomObject/PSObject is
You can add new methods and properties to PSCustomObject/PSObject
You can use PSCustomObject/PSObject for pipeline parameter binding using ValueFromPipelineByPropertyName as explained by Zombo
example: [PSCustomObject] #{Path='foo'; Type='directory'} | New-Item