Medini 2022+ Scripting and API Documentation

Question

We have recently upgraded to medini 2023 from medini 2021 and noticed a few changes that were not communicated sufficiently. A minor observation was that medini 2023 was stricter on variable declarations - this is easily understood and a good practice to update anyway.

A few not so minor corrections we had to make include how constructors are handled:

/*OLD*/ if (arg_0z.constructor === Metamodel.safetygoals.SafetyGoal) {}
/*NEW*/ if (Object(arg_0z) == Metamodel.safetygoals.SafetyGoal) {}

"sysml_port" - Metamodel.sysml.SysMLFlowPortUsage and "sysml_port_write" - Metamodel.sysml.SysMLPortUsage object meta class changes (mediniGetContainer returns sysml_port_write now and was sysml_port before):

/*OLD*/ if (medini_container === "sysml_port") {}
/*NEW*/ if ((medini_container === "sysml_port") || (medini_container === "sysml_port_write")) {} //make backwards compatible

There may also be a change in how sysml objects return children and the use of direction attribute may be needed to verify between "sysml_in_port" and "sysml_out_port".

In general, we would like to be aware of the details of these changes as we upgrade medini versions in the future...

medini_1024169437 · Answer

Thanks Johan for the clarification.

Point 2 and 3 are clear now. About point 1: we did not change that explicitly. There is a new option in the prefereces (Scripting) that allows you to enable "strict" mode always without changing any script.

Maybe that is enabled in your case? Its off by default though.

Another thought: we have changed a bit the way how scripts are executed as mentioned and it *might* be that one strict script is loading or executing another script that was not stcut compliant before.

"Implicit" means for example, a batch script is makred "strict" and it uses a derived scripted property that is not strict but now executed in the context of the barch script.

Hope that helps to clarify the behavior a bit.

Happy Scripting

Jan

medini_1040753747 · Answer

Prior to medini 2022, it was allowing missing "var" before. Example of stricter variable declarations:

/*OLD*/ my_var = 0;
/*NEW*/ var my_var = 0;

Didn't try to use "instanceof" before - that seems to be a good solution as well!

When building models we used Metamodel.sysml.SysMLFlowPortUsage before now it seems to be using Metamodel.sysml.SysMLPortUsage. Sorry for the confusion... the previous example was using nicknames we converted to on our side.

medini_1024169437 · Answer

Hi Johan,

thanks first of all for your findings. In general I think there might be two reasons for that:

* Our constant attempt to move our sysml metamodel to sysml v2
* Some defects we fixed in how the JS engine is initialized and handling Java objects, mainly to make BIRt scripting and batch JS scripting more similar and to fix defects in BIRT scripting.

For the first observation (medini is more stricter) - d you have an example for that? As we did not really change something here.
The second point is a bit unclear to me. obj.constructor is not code we directly support, I am not sure what it is supposed to do. In the past we supported obj.prototype which returns the metaclass and can be compared to Metamodel.something but that was later improved by supporting instanceof. Can you give some insights into that code?
The third finding: how is medini_container defined, could you write that shortly? I want to understand where this "write" suffix is coming from.
Happy ScriptingJan