The attached sheet used to work and now doesn't.
The regression happened somewhere between build 7803 (works) and 7822 (doesn't work).
Also strange: the german error message even though the UI language was set to english.

Vektoriteration AEWP.sm (17 КиБ) скачан 64 раз(а).

Confirmed: works fine SS 6179

Is this the correct result?

Yes, it is correct. Yet I consider the need to place eval() being a workaround for a bug. A similar workaround is to introduce an intermediate variable.

It seems that the error is related with broken logics of matrix operations. If I remove the factor in front of the matrix, then the function call works. Seems that at some point the evaluation logic thinks that the expression is not a matrix and complains in a not very helpful way.

Actually I found the culprit, it's the Ones() function. Davide, if you could ask that function to evaluate it's argument before refusing it, that would be fine.

Wrote

1) Yet I consider the need to place eval() being a workaround for a bug.
2) Actually I found the culprit, it's the Ones() function.

1) I agree, putting line() or eval() sometimes is cumbersome. And unnecessary.
If necessity of them would be removed in all possibilities, it would be awesome.

2) You are 100% correct. I wonder how many of these little bugs out there.
This little change of code worked just fine.


Regards