Introduce forever time literal

[Depetrol]

This PR introduces a new forever time literal that can be used as STA and STAA in decentralized execution. This PR is related to #2368 and closes #2374.

[lhstrh]

Keywords in LF are all lowercase -- let's stick with that?

[Depetrol]

Makes sense! I'll modify the PR.

[edwardalee]

It's unclear to me how this is used. I don't see any changes in the code generator to detect "forever" and then use the C macro FOREVER. How does this get used?

[Depetrol]

This implementation does not use the C macro FOREVER. Instead, it introduces a new Lingua Franca time literal named forever in the Xtext syntax, and when the compiler sees this time literal it interprets this time literal as TimeValue.MAX_VALUE. This is similar to how the compiler currently handles the 0 time literal. The benefit of this implementation is that it is target language agnostic.

[edwardalee]

This implementation does not use the C macro FOREVER. Instead, it introduces a new Lingua Franca time literal named forever in the Xtext syntax, and when the compiler sees this time literal it interprets this time literal as TimeValue.MAX_VALUE. This is similar to how the compiler currently handles the 0 time literal. The benefit of this implementation is that it is target language agnostic.

Ah yes, that makes sense.

[edwardalee]

This looks good to me.

[lhstrh]

Is there no practical use case fornever? If there is, then shouldn't it also be allowed as a time value?

[Depetrol]

I haven't yet encountered a use case where never is used. However, if there is a use case for it, I would be glad to add it to the syntax as well!

[Depetrol]

I was looking into the tests and found a test LastTimeReplace.lf that uses the never time literal. It is noted that

reaction(startup) {=
    // Unfortunately, a time state variable cannot be initialized with NEVER.
    // So we do that here.
    self->last = NEVER;
  =}

I think introducing the never time literal makes sense in this use case.