Experimental Physics and
Industrial Control System

Michael Davidsaver via Tech-talk <tech-talk at aps.anl.gov> · Thu, 18 Jun 2020 13:42:51 -0700

On 6/18/20 9:18 AM, Ralph Lange via Tech-talk wrote:
> On Thu, 18 Jun 2020 at 17:37, Michael Davidsaver <mdavidsaver at gmail.com <mailto:mdavidsaver at gmail.com>> wrote:
> 
> 
>     Why not bake procServ into the IOC executable?  It could still fork() off the a child process for the actual IOC, but without an execvp().
>     (or at worst, to re-exec the same executable)
> 
>     Of course fork() from a multi-threaded process needs some care.
>     As well as attention to FD_CLOEXEC.
> 
> 
> As you know, the traditional Linux approach is to break up applications with large complex functionality into small independent units that each serve a narrow scope of purpose and can be combined on the command line.
> procServ is following that approach.

Quite true.  And imo this should continue to be an option.
I guess I'm just not such an orthodox UNIX head to see this
as a reason not to try a new idea.

> Looking at the list of command line options and features of procServ (that were all added on user request btw.) - do you want all this being part of the IOC?

Those that apply.  Sure.

> Your idea sounds a lot like the CA Gateway's quirky server mode - that IMHO should be deprecated in favor of running it under procServ, to put an end to running two hardly distinguishable processes for each Gateway instance.

Ok, it was only a passing thought.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System