You are on the right track. I suggest to continue focusing on the code of the rule itself. Don't worry about the rest like docs or some other test failures at the moment, they will be easy to fix at the end.

Christoph Bergmeister (@bergmeister) I've hit a milestone. At this point the code seems to be working as intended. I have added one minor feature which may need to be in a separate PR: I've added the option to use Test-ScriptAnalyzer with a -RuleToTest parameter, so I could easily make adjustments, run .\build.ps1, and run Test-ScriptAnalyzer -RuleToTest 'avoiduni' -ShowAll to check if the change made it worse or better :-).

Can we have a little brainstorm on where this rule may be applied to, other than just for Foreach-Object -Parallel? I'm thinking of maybe Invoke-Command, but other than that? [Edit: a list of commands where this could be applied to can be found here]

Would you like to start a review round at this point, or better after we have a clear picture of how to move forward?

The failure in Ubuntu 1804 build is the same Authenticode glitch we saw earlier.

// brainstorm

Other situations where this rule can be applied:

• Start-Job/Start-ThreadJob -scriptBlock {$using:foo}
  Here the catch is, we need to be sure we check the right scriptblock. The rule does not apply to the -InitializationScript scriptblock parameter.

• Invoke-Command -ComputerName "baz" -ScriptBlock {$using:foo}
  The catch here, is that the -ComputerName parameter has to be used, otherwise $using is prohibited.
  Also tricky; one can open a persistent session, and use subsequent invoke-command calls to that, where variables previous calls are still valid (more info here)

  $session = new-pssession -computername "baz"
  Invoke-Command -session $session -scriptblock {$foo = "foo" }
  Invoke-Command -session $session -scriptblock {Write-Output $foo} # works

• workflow bar {$foo = 'foo'; inlinescript {$using:foo} }
  On Windows PowerShell only. (see get-help about_InlineScript)

Michael Klement (@mklement0) thanks for creating the powershell-docs PR for documenting this. That gave me the quick overview I needed. If you have any suggestions, I'd love to hear them, thanks.

Thanks, Jos Koelewijn (@Jawz84), a great rule to add, and it would be great if it covered the other $using: contexts too (as you outline above).

One pending PR to watch out for: PowerShell/PowerShell#11829, which is trying to eliminate overzealous $using: checks for nested script blocks, as described in PowerShell/PowerShell#11817

As for the terminology in the rule documentation: about_Remote_Variables refers to $using: as the Using scope modifier - it's not a directive.

Thanks for pointing these out Michael Klement (@mklement0) !

The rule is working now. I am not completely satisfied with what the implementation looks like, and I'm likely to do some cleanup and refactor. If Michael Klement (@mklement0) or Christoph Bergmeister (@bergmeister) have insights how to best convert this to cleaner code, I'm all ears.

Second question: when starting out, I named the rule AvoidUnInitializedVarsInNewRunspaces Is that the best, most descriptive name for this rule?

Rob Holt (@rjmholt) I've been thinking about what to optimize, and especially: how to measure if it has worked. When I run tests, most are around 15ms, but some stand out and take longer:

I have tried Visual Studios profiling tool and the diagnostic tools, but none the wiser really. How do I visualise which codepaths make the Non-DSC function with the name SetScript {} test take 156ms instead of the more typical 15ms?

Rob Holt (@rjmholt) did you have a chance to look at this yet? I'm starting to fear we might be accidentally waiting for each other

Really enjoying your feedback Rob Holt (@rjmholt)! I am learning a lot.

Rob Holt (@rjmholt) back over to you, all good here.

Christoph Bergmeister (@bergmeister), Rob Holt (@rjmholt) thanks for all the reviews. Is there anything more I can do here? I am uncertain what the next step is. Is this blocked on review from James Truher (@JamesWTruher)?
Thanks in advance for enlightening me 🙂

I think having at least one reviewer from Microsoft is good enough, this is our unwritten rule, so I'll therefore merge. Congrats and thanks for your hard work