Re: [Bug-apl] Scan operator has strange behaviour

[Elias Mårtenson]

On 19 March 2014 22:25, Kacper Gutowski <address@hidden> wrote:

On 2014-03-19 21:22:04, Elias Mårtenson wrote:
> This can't possibly be correct?

But it is!
Let me just quote relevant excerpt of ISO 13751:

OK, I understand your explanation, and I have to say that the spec seems to be a bit over-specified here. My (simplistic?) understanding of the backslash operator was that it simply applied the given function on the first two elements, add the result to the list, and then repeat the process with the result of the previous iteration and the next element in the list.

The spec talks about associativity of the function, but given the current reading I can't see how associativity would make any difference (it wouldn't given my understanding above, either). To me, the sentence only makes sense if the work "associativity" is replaced with "side-effect free".

Are there any cases where a side-effect free (but non-associative) function applied to the backslash operator would give different results?

Regards,

Elias