bpo-36638: fix WindowsLoadTracker exception on small windows SKUs

[paulmon]

typeperf.exe is not present on small editions
of windows like Windows IoT Core or nanoserver
This causes WindowsLoadTracker to throw an exception during test initialization.

@ammaraskar @zooba

https://bugs.python.org/issue36638

[ammaraskar]

Whoops, I assumed typeperf would be available on all supported Windows editions. I think this fix might leave some open handles around, the pipe_write_end and self.pipe in win_utils but this might not be a big concern since its just for the duration of the test suite run.

Aside from that, looks good to me.

[zooba]

I'd rather expand the platform check to exclude ARM. That way we'll find out about changes here in the future faster (when it breaks, rather than when we notice a warning).

[paulmon]

This isn't an ARM issue.
typeperf.exe is not available on Windows IoT Core x64, or nanoserver x64 either.

It makes more sense to check for the existence of c:\windows\system32\typeperf.exe, or typeperf.exe on the path, than to limit it to arm. I'm not sure which might be more pythonic. The missing dependency is what is causing the exception, not the processor architecture.

[zooba]

In that case, handling the exception when it's not found is better. Sorry to send you back to the first commit (but fix up the formatting while you're there - no space between function call (print) and its arguments, and capitalizing product names is totally okay)

[paulmon]

No problem. I'm happy to learn.

Alos, are there any VS Code addins or other tools that will help me get the whitespace right before I create the PR?

[ammaraskar]

Using a linter like pylint will catch trailing whitespace issues like those https://code.visualstudio.com/docs/python/linting (Note that some of the stdlib doesn't conform to pep8, so you might run into some false positives)

If you don't want to use a full blown linter, for that particular issue you could use something like https://marketplace.visualstudio.com/items?itemName=shardulm94.trailing-spaces

[ammaraskar]

The str() call here is redudant, an empty format specifier is essentially the same as calling str()

A general convention is that an empty format string ("") produces the same result as if you
had called str() on the value. A non-empty format string typically modifies the result.

https://docs.python.org/3/library/string.html#format-specification-mini-language

[paulmon]

Thanks!