Rewrite OMCSessionZMQ

[syntron]

This is now a complete rewrite of OMCSessionZMQ; it is defined on top of PR #300

The main items are still the same as mentioned in the original statement.

Besides the OMCSessionZMQ class there are now several classes which can be used to create the OMC/ZMQ process:

• OMCProcessPort
• OMCProcessLocal
• OMCProcessDocker
• OMCProcessDockerContainer

This will also allow usage of WSL (see #268); I have some code ready but could not test it ...

---

OMCSessionZMQ handles different tasks:

a) start OMC process (local / via docker / via docker container)
b) connect to OMC process via ZMQ
c) connect to OMC using ZMQ

This PR tries to find a way to to split these items into several classes:

[1] OMCSessionZMQ - for b) and c)
[2] OMCProcess / OMCProcessLocal / OMCProcessDummy / OMCProcessDocker(?) - for a)

I started to work on this due to the docker code just sitting there without much usage in my case. Save for docker, all other test run fine ...

Still work in progress; feel free to comment if this even makes sense ...

On top of PR #291

[syntron]

@adeas31 I just tried to run OMPython with docker ... and it failed even for 3.6.0! At that time docker was last tested? Who is using it?

PS: got it working again ... it depends on omc installed on the host (in 3.6.0 and also now)

[syntron]

@adeas31 this is now a complete rewrite of the code to connect to / start OMC in different ways:

(1) just provide the ZMQ port => OMCProcessPort
(2) local OMC => OMCProcessLocal
(3) in docker => OMCProcessDocker
(4) via docker container ID => OMCProcessDockerContainer

I think, what it would be quite easy to add podman or similar tools based on this definition. For the default users (localy started OMC), nothing changes.

Question: is this something for the next release? If yes, I would cleanup this code and prepare it as PR ... however, more testing would be needed! I can only test a limited set of possibilities ... port (win/linux) / local (this is tested on github) / docker (linux) / docker container (linux)

[syntron]

@adeas31 Please test and consider for the next release - it is quite a big change! (see #254)

[adeas31]

LGTM

[adeas31]

Perhaps you can define a proper constructor for OMCProcessDockerHelper which sets the docker arguments using the super().__init__ instead of setting them explicitly in OMCProcessDocker and OMCProcessDockerContainer.

super().__init__ can handle it based on cooperative multiple inheritance.

[syntron]

I have some updates for this series including the requested change; however, I'm not sure if it would be good to put all into one PR. I prepared it as PR #306 (includes #301)

[adeas31]

Great. Please rebase the PR and we are good to merge it.