Question about interconnection of terminals

[masoud-najafi]

HI,
I have a few questions regarding the interconnection of terminals.

1. Specifically, can two terminals with different matchingRules be connected? Unless I've missed something, my interpretation is that the matchingRule should be checked independently for each terminal. For example:
   A plug bus terminal (Tp) can connect to a bus terminal (Tb) as long as all memberNames of Tp are present in Tb.
   It is not required for the memberNames of Tb to also exist in Tp. Is this interpretation correct?

2. How should matching of two nested (hierarchical) terminals be checked? Specifically, should the memberNames of nested terminals be unrolled and flattened in depth-first order then matching rules be checked? In other words, if a terminal contains hierarchical sub-terminals, does the connection logic require recursively traversing and flattening the structure? Is depth-first traversal the expected method for resolving memberNames in such cases?

3)Can a single variableName (a variable from ModelVariables of an FMU) be shared across multiple terminals of the same FMU?
Thanks,
Masoud

[chrbertsch]

Pierre: for Terminals we start going into the area of "what can be connected". We give more heuristics of what can be done and what not ... This should be read as heuristics ... For a "bus" e.g. you should not worry if you cannot connect everything
Andreas: ... I think the behaviour is defined
Pierre: in a given context e.g. with Modelica tools and MSL it is better defined.
Pierre: Terminal help users to connect FMUs ... will work best within a narrow user community.
Andreas: We do not want to restrict usage of FMUs beyond reasonable constraints.

Pierre: 1) is a reasonable interpretation
3) yes
Pierre: 2) is not obvious for me.
Klaus: Terminal can be ignored ...

Andreas: I suggest not to answer the questions in detail. OK?
Klaus: it was the original idea to create layered standards for this. E.g. for the MSL.
Elmir: for the causality in Modelica you need to make assumptions.
Pierre: define a) how to map to terminals b) how to fix causalities
Pierre: We should create a terminals sandbox.

We could add a non normative text that terminals are heuristics.

[chrbertsch]

We created a sandbox to share FMUs with terminal (can also be from prototype implementations):
https://github.com/modelica/fmi-terminals-sandbox/