[R-pkg-devel] automated testing for an SSH tunneling package?

[Bob Rudis]

I'm not sure where Jeroen is on this - https://github.com/jeroenooms/ssh -
but it might make more sense to dovetail off of it than rely on binaries
being available on systems. That's doable, but (IMO) fraught with peril.

On Fri, Sep 16, 2016 at 4:53 PM, William May <williamcmay at live.com> wrote:

> I wrote an SSH tunneling package (it's on Github here<
> https://github.com/wmay/tunnelr>). It runs on Mac, Linux, and Windows, at
> least in the environments I've been able to test. I would like to get it on
> CRAN if it's possible, but writing automated tests for this package is
> nightmarish.
>
> Typically, connecting to a server through the package requires manually
> entering a password, which I don't think I can automate. (I don't want to
> allow the package to enter passwords itself for security reasons.) An
> alternative is to set up public and private rsa keys so that no password is
> required. That option also gets complicated.
>
> I could imagine setting up a server solely for testing this, with a user
> that has no password, though that sounds scary. But then what happens if
> the server goes down?
>
> A second challenge applies specifically to Windows, which is the platform
> I'm most concerned about. I'm relying on PuTTY for SSH in Windows. Is there
> a way for that to be tested on CRAN?
>
> Should I even worry about CRAN testing for this? Will CRAN accept this
> package without tests?
>
> And I was surprised that an SSH tunneling package didn't already exist (I
> found it really convenient to control SSH tunneling from within R). Is
> there another reason it doesn't exist yet, or is it just the testing
> challenges?
>
> - Will
>
>         [[alternative HTML version deleted]]
>
> ______________________________________________
> R-package-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-package-devel
>

	[[alternative HTML version deleted]]