mdavidsaver

Ah, yes. Move the nameserver behind the gateway... This seems like it should work so long as there is only one level of gateway.

imo. any re-working should favor [`std::map`](https://en.cppreference.com/w/cpp/container/map) as the more sustainable, and perhaps better performing, alternative. Maybe with `ifdef`s to use [`std::unordered_map`](https://en.cppreference.com/w/cpp/container/unordered_map) when available.

I take it that merging https://github.com/epics-modules/pcas/pull/3 is a prerequisite?

What is the story for concurrent access to the new `defunct` and `shutdownNeeded` flags? I have almost convinced myself that all changes are made under the port lock, but I...

> ... However, having three mutexes involved gives me a headache, so it might take a while. Not a request, only a thought. I've taken to using comments within struct/class...

> However static builds of those downstream applications won't necessarily know about tirpc ... Yup, I know. I see this PR as a first small step towards dealing with a...

I am also thinking about ways to give a better error when `rpc/rpc.h` is missing. Something which might avoid some of the tech-talk traffic. eg. something like ```c #ifdef __has_include...

Also, I feel bound to point out that this PR, and any which may follow, are in effect re-inventing functionality which is present in all ~modern C/C++ build tools. Anything...

> (Do we have a #define for "the new network stack" ? I don't believe that the RTEMS devs. have an "official" solution. Practically, there are a number of headers...

For that matter, these commands could also be taught about `iocshSetError()` (from Base 7.0.3.1) to be able to participate in IOC shell error handle with the `on` command (eg. `on...