I am writing a code to run a pi-based robot. It has several sensors and communicates with external computers over wifi as well. Some of the sensor and computer communication data is done over sockets (to other scripts) to keep things modular and simple (for me). As long as I stay at less than 10 sockets, will I come across any problems? I am mostly wondering if there is anything inherently wrong with this coding strategy.
Q : As long as I stay at less than 10 sockets, will I come across any problems?
Maybe yes, maybe not. The number of ZeroMQ Socket-instances does not matter in this, the configuration of the ZeroMQ Context()-instances and thereof of the underlying operating system's resources do matter.
Q : anything inherently wrong with this coding strategy?
No, unless you do right things in a wrong way. Particularly "this coding" remains magically hidden as an unseen, so far, code, the less the MCVE-formulated code-example and the general answer for such a problem thus suffers from a principal undecidability, until an MCVE-code is let to run in the problem-specific eco-system, with known limits for the control-loops and the robot physical properties ( which is left to the kind readers to review ad-hoc ).
ZeroMQ will have Zero-problems in running on the RPi-platform, given due configuration does not suffocate the Context()-instance(s) by not providing adequate resources.