TLC doesn't like multiple types on function range

[JosEdu Solsona]

Hello,

Is it possible to define a function with a multiple "type" range like [Nat -> Nat \union BOOLEAN] without TLC complaining about the type?.

When checking a type invariant will result in:

Attempted to check if the value:
FALSE
is in the integer interval 0..2

If "{a,b}" are untyped model values then there is no problem with [Nat -> Nat \union {a,b}], since "a" and "b" are by definition
unequal to anything but itself. But in the case of BOOLEAN, TLC refuses to compare with anything of different type.

Thanks!

[Stephan Merz]

The semantics of TLA+ doesn't specify if FALSE = 0 [1], for example, so TLC is rightly refusing to evaluate such expressions, and it is hard to see why it would be useful to use a function of type [Nat -> Nat \union BOOLEAN] in a specification.

Regards,

Stephan

[1] Note that this identity is true by convention in C.

--
You received this message because you are subscribed to the Google Groups "tlaplus" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tlaplus+u...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tlaplus/a7fc436e-e2d6-4527-882e-3dcae49c8716o%40googlegroups.com.