ja11sop

This is still of interest to me - I see most of the code relating to this has changed - I'll see what I can do

I'm looking at this now - it's still broken but the fix might be minor.

The current PR has no conflicts and supports the use case of calling init() on the framework more than once which we use to allow us to launch test suites...

You mean [https://conan.io/](https://conan.io/) for anyone who stumbles across this and isn't sure what Conan is. I'll look into how some sort of support might be added for Conan. Though I...

Yes - good link. I'm not sure what's the best approach here though. Is this something we fix or advise a work-around for as a "better" practice. For instance the...

I don't have a lot of experience with using `shell=True` - are the warnings around the use of this argument relevant to our use-case here? Should this be configurable or...

Sorry for the long delay in responding - I haven't gone ahead with this merge yet as adding `shell=True` seems to break on Linux - at least with my test...

So what are the options? Remove it or make it equal to a v4 address?

It seems that Option 1 has the advantage of being known to work well. The other options seem to open the door to possible side-effects yet to be well understood.

Thinking about Option 4 I could see how allowing a 'not an address' state could lead to lots of validity checking in user code that takes an `ip::address` either pre-emptively...