Reference equality on numbers that aren't non-small integers

[ltratt]

AIUI, = means "perform equality check on an object's data" and == means "pointer equality". However the latter is overriden for small integers, but not big integers or doubles. So, given this program:

obj1 = (
    run = (
        (1 == 1) println.
        (100 == 100) println.
        ((1 << 200) == (1 << 200)) println.
        (1.0 == 1.0) println.
    )
)

the output in java-som is:

true
true
false
false

In the interests of consistency, I tend to think those should all be true or all be false. If they should all be false, then one has trickier cases like 1 == 1.0 and so on. Perhaps reference equality should always return false for numbers?

[smarr]

For integers, the test suite prefers that == works on all value ranges:
https://github.com/SOM-st/SOM/blob/master/TestSuite/IntegerTest.som#L30-L53

Thus, I'd consider this "specified", and (1 << 200) == (1 << 200) should be true.

Though for Double, there aren't any tests.

Historically, pointer equality was always a point where one was able to observe implementation strategy.
Form a language design perspective, I'd argue, one wants to avoid such leaky abstractions, and 1.0 == 1.0 should be true.

For 1 == 1.0, I actually argue that it should be false. These are clearly not the same objects.
Returning always false seems like a legal but lazy hack.

Unrelated, there is a strong precedence and preference for class names alway to start with a capital letter. obj1 really needs to be Obj1.

[ltratt]

Thus, I'd consider this "specified", and (1 << 200) == (1 << 200) should be true.

Makes sense.

For 1 == 1.0, I actually argue that it should be false. These are clearly not the same objects.

OK, fair enough, I can certainly live with that.

I'll adjust yksom accordingly.