VMs are useful to improve functional safety and "security", isolating function software components versus each other. Given some hardware support (especially an MMU) is available, current "big" MCUs come well-prepared for that. And it is - with varying denominations - already a trend in eg. automotive ECUs.

What I do not see is the need for some DSL (interpreting this as "Domain Specific Language"): VMs would be well able to cope with native binary code - just adding some "system abstraction layer" on which the application is built.

For more information

there are many types of programs taken in input of an embedded VM, depending from that VM:

In a general way, I think that the use of a VM is justified in a specific purpose, i.e. Domain Specific

Hello Gerard,

most of your examples describe interpreters. (One might discuss whether this applies to your 3rd topic as PLC programs usually result in compiled code.)

Regarding what the abbreviation "VM" is covering (see https://en.wikipedia.org/wiki/Virtual_machine ), an interpreter is way less than a VM. Maybe you want to start a new thread?

Regards

Exactly, I want to start a new thread; but I want to make sure that it could be usefull.

I have proposed a mathematical representation of models written in Grafcet language, and for its realization, it is convenient to deploy on µCs an interpreter. I am thinking about advantages and disadvantages of such a solution.

Thanks

From the application view, it would be better if the code were compiled: typically shorter and way faster.

If you're "searching" something for a paper or thesis: instead of targeting some interpreter, some "code generator" ("translating" Grafcet to eg. C/C++ - subsequently compiled by the target system's compiler) would be of similar complexity while resulting in better performance. If you want to add complexity: target-independent code generation would be a plus.

This approach would be within the scope of current trends.

Regards

Thanks very much