Windows testing hangs

[gowerc]

On some windows machines testing the package via R CMD check hangs indefinitely. Originally I had thought this was just a problem with the windows runners for github actions however I was able to re-create the issue on our internal windows VMs:

Interestingly its not all windows builds for example rhub's windows machine and CRANs win builder do not have this issue. Unfortunately the VM I have access to is so under powered and restricted that its incredibly hard to diagnose this.

[gowerc]

@nociale - when you get a spare moment would you be able to see if you can re-create this ?

Essentially on the windows terminal you first need to build and check the package via the direct R commands e.g.

cd <path/to/rbmi>
R CMD build .
R CMD check .\rbmi_1.2.4.tar.gz

On the VM this would then hang forever after it reached Running 'testthat.R'

[gowerc]

@nociale / @wolbersm - After talking to @danielinteractive we were considering dropping support for parallelisation on Windows machines (assuming that is the root cause of this issue).

We are making the assumption that the vast majority of users who are using this package in production who need the ability to scale their compute will be using Linux servers and thus (assuming parallelisation is the cause of our issues here) that there is no real cost to just dropping support for windows in order to resolve this bug.

From an engineering perspective we are currently having to do a lot of significant hacky workarounds to get parallelisation to work on Windows so it would greatly simplify our code and maintenance if we don't have to support it.

[wolbersm]

Fine by me.

[gowerc]

Interestingly in the current branch that implements Stan as a suggested dependency instead of a import the issue no longer appears to occur. Fingers crossed that this might have been accidentally fixed.