Possible discrepancy in the documentation regarding the use of script variables

[ronaldborman]

Regarding the use of script variables, in the third recommended pattern in Outputs and Artifacts the documentation says:

These variables are only accessible within the same Invoke-psake call.

But the last paragraph of Variable Scope and References says:

In case you create a variable with script scope, the variable is kept inside the psake module. This means, that if you run Invoke-psake again, the variable is available.

I've did some tests and it looks like a script variable is only available in the same invocation. Is the second statement incorrect or am I missing something?

[HeyItsGilbert]

Good catch. Thanks for making the discussion. When $script: scoped variables are set they become available to the module. I think with 5 we changed how the invocation happens and the last paragraph may no longer be true. Let me consider some tests to add and run and I'll get the docs updated. Thanks again!

[HeyItsGilbert]

After digging into the v5 source and running some tests, I have an update — I was wrong in my initial reply.

$script: variables do persist across Invoke-psake calls in v5 which matches my mental model. Here's what the tests showed:

Test 1 — build file with a Properties block:

• Call 1 runs SetVar, sets $script:crossCallVar = "set_by_first_call"
• Call 2 runs ReadVar, prints "reset_by_properties"

Looks like it didn't persist but it did. The Properties block ran before ReadVar and overwrote the value.

Test 2 — build file without a Properties block:

• Call 1 runs SetVar, sets $script:bareVar = "bare_set_by_first_call"
• Call 2 runs ReadVar, prints "bare_set_by_first_call" ✓

The variable survived the second invocation untouched.

Verdict: The Variable Scope doc's last paragraph is correct. The inaccuracy is in Outputs and Artifacts, which says $script: variables are "only accessible within the same Invoke-psake call" — they're not; they live in the module scope and survive across calls.

The real gotcha worth documenting: if you declare the variable in a Properties block, that block re-runs on every Invoke-psake call and will silently overwrite whatever value the previous call left behind — making it appear like the variable didn't persist.

I'll update the Outputs & Artifacts doc to correct that claim and add a note about the Properties-block behavior.

[HeyItsGilbert]

psake/docs#56 for the docs update.

[ronaldborman]

Thanks for the investigation and clarification. The Properties block indeed tripped me up when I tested it.