May 07 2020 12:18 PM
Why do IsEven() and IsOdd() give a #VALUE error when used with a dynamic array parameter? E.g.:
A1: =Sequence(10)
B1: =IsEven(A1#)
C1: =IsEven(--A1#)
D1: =IsNumber(A1#)
In the example above, B1 gives a #VALUE error, while C1 works as expected. (Using NumberValue() instead of "--" also works.) Interestingly, D1 works as well. In fact, all of the Is<x>() functions play nicely with dynamic arrays, with the exception of IsEven() and IsOdd().* This, plus the fact that they work nicely in conjunction with double negative and/or NumberValue() implies that this is a bug. Is that correct?
* IsFormula() and IsRef() are unusual cases. IsFormula(A1#) does output a dynamic array, although the result isn't what I would expect, and I can't think of a good reason it would be implemented that way, but so be it. IsRef(A1#) outputs a scalar value of TRUE, which is somewhat understandable, although it would seem more consistent to have it output a dynamic array and instead have people use IsRef(A1) to output a scalar value. On a somewhat related note, I wonder if there are plans for a version of Type() that will map to array elements and output a dynamic array? Perhaps as an optional second parameter?
Thanks,
-Jeremy
May 07 2020 12:30 PM
Also interestingly, IsEven({1, 2, 3, 4}) works, as does IsEven(Sequence(4)). IsEven(A1:A4) gives a #VALUE error, while IsEven(--A1:A4) works.
It seems like IsEven() and IsOdd() work with arrays, but not with references to arrays or ranges. As mentioned above, all of the other Is<X>() functions work with references to dynamic arrays or with ranges. This definitely seems like a bug, or at least a completely unnecessary functional inconsistency if this is somehow the desired behavior.
Thanks,
-Jeremy
May 07 2020 01:00 PM
Jeremy, not only. For example =EOMONTH(TODAY(),A1:A4) returns an error and =EOMONTH(TODAY(),+A1:A4) returns the spill. Some other functions also do not work with ranges and arrays in a same way.
We may consider that as a bug since we expect bit different functionality. Or we may consider that as a different behaviour, specific for different functions.
In any case I don't think old functions will be re-writing to make them work exactly the same way with dynamic arrays. It take years and introduces a lot of new bugs. Is not worth.
May 07 2020 01:06 PM
Trying to find an explanation.
IS functions states every IS function except ISODD() and ISEVEN(). And is states that a "value" is required.
ISODD() on the other hand requires a "number".
May 07 2020 05:57 PM
I get that IsEven() and IsOdd() require input values to be numbers, while IsNumber() & other IS functions don't. But the input parameter is an array of numbers. IsNumber(A1#) returns an array of TRUE values. More to the point, IsEven(Sequence(10)) works, while IsEven(A1#) doesn't. And IsEven(+A1#) does! This is definitely a bug. There's something about a reference to a dynamic array that "poisons" the input for IsEven()/IsOdd(), where giving it the underlying dynamic array directly and/or "cleaning" it with any other function ("+" or "--") works.
So IsEven() and IsOdd() can handle dynamic arrays just fine, but not through a direct reference. That's not right - completely anathema to how Excel works. The entire product is about composing formulas by referencing other cells (which contain formulas)!
-Jeremy
May 07 2020 09:42 PM
Based on the support pages I suppose that ISODD()/ISEVEN() are programmed differently then the other IS functions.
ISEVEN() expects a number or a range of numbers. That is why +A1# or SEQUENCE() inside ISEVEN() work. However A1# is not recognized as a (spilled) range. Not really a bug because ISEVEN() was there before spilled ranges but definitely an inconsistency that should be fixed by Microsoft.
May 08 2020 08:19 AM
Potato potato - agreed. :) I'm not aware of any other function where F(expr) works but F(ref/range), cell/range=expr doesn't, with the possible exception of functions explicitly working on type/cell metadata - e.g., ROW(), TYPE(), ...