Get rid of dependency on deprecated WMIC on Windows

[janbuchar]

• We depend on WMIC via https://github.com/apify/ps-tree/blob/master/index.js, called from LocalEventManager to get memory usage (or cpu? doesn't matter...)
• https://learn.microsoft.com/en-us/windows/win32/wmisdk/wmic
• It doesn't matter if we resolve it by fixing our ps tree fork or by using something else
• pointed out by @nicolatournay in Error 4058: Can't even run the Getting started example #2764

[NathanSavageKaimai]

It appears the current implementation is broken anyway. If you call getMemoryInfo on windows, the returned object returns a default value for the mainProcessBytes

{
  totalBytes: 25438089216,
  freeBytes: 7435272192,
  usedBytes: 18002817024,
  mainProcessBytes: -1,
  childProcessesBytes: 14073856
}

Im not quite sure how this has gone unoticed, the test suite for it fails because of this. Is the test suite not regularly run on windows?

[janbuchar]

@NathanSavageKaimai thanks for bringing this up, we really should run the unit tests on Windows as well.

[NathanSavageKaimai]

@janbuchar ive been playing around with this and I have a solution that works but ive hit a bit of a snag. The modern alternative to WMIC that Microsoft reccomends is the powershell command Get-CimInstance. This can be executed as follows:

            processLister = spawn('powershell', [
                '-NoProfile',
                '-Command',
                'Get-CimInstance Win32_Process | Format-Table ProcessId,ParentProcessId,WorkingSetSize,Name',
            ]);

unfortunately this takes about 66% longer to execute than WMIC did. ~500+ms vs ~300+ms. Is this a trade off you can accept or should I keep looking?

Theres a module i found called ps-list by sindresorhus that has a custom binary that would likley be faster but obviously it would add a third party dependancy.

If the powershell solution is acceptable, a slight test modification is needed in the test suite as the Snapshotter ".get[.*]Sample limits amount of samples" test doesnt run long enough for the function to run twice and thus throws a false positive.

Let me know what you think. : )

[NathanSavageKaimai]

this issue should now be resolved but only when using the systemInfoV2 configuration flag