Cycle enumerator

Hi,

I was surprised today with a behaviour, maybe my expectation was wrong
and someone can explain why it works this way:

I have an array of elements, and I want to cycle through it assigning
each element to another list of elements. But the cycle can be already
in the middle. For example:

a = [1,2,3]

b = [6,7,8,9,10]

the desired result should be (if the next element in the cycle was 2):

[[6,2], [7,3], [8,1],[9,2],[10,3]]

So I, intuitively tried this:

2.0.0-p195 :469 > b = [6,7,8,9,10]
=> [6, 7, 8, 9, 10]
2.0.0-p195 :471 > a = [1,2,3].cycle
=> #<Enumerator: [1, 2, 3]:cycle>
2.0.0-p195 :472 > a.next
=> 1
2.0.0-p195 :473 > a.peek
=> 2
2.0.0-p195 :474 > b.zip(a)
=> [[6, 1], [7, 2], [8, 3], [9, 1], [10, 2]]

So, it seems that the enumerator a is starting from the beginning,
instead of from the "next". If I read correctly the source code (MRI)
it seems that zip uses take, and testing this shows that it always
starts from the beginning:

2.0.0-p195 :475 > a.take 2
=> [1, 2]
2.0.0-p195 :476 > a.take 2
=> [1, 2]
2.0.0-p195 :477 > a.take 2
=> [1, 2]

Is it wrong to expect the enumerator to behave as a cycle starting
from its current position for operations like zip and take?

Jesus.

This is surprising behavior to me. In case no one finds a good explanation
in this mailing list, I think you should send a ticket to ruby-core asking
why it works this way (and fixing it in case there isn't a good reason to).

···

-----
Carlos Agarie
Software Engineer
+55 11 97320-3878 | @carlos_agarie

2014-06-26 13:21 GMT-03:00 Jesús Gabriel y Galán <jgabrielygalan@gmail.com>:

Hi,

I was surprised today with a behaviour, maybe my expectation was wrong
and someone can explain why it works this way:

I have an array of elements, and I want to cycle through it assigning
each element to another list of elements. But the cycle can be already
in the middle. For example:

a = [1,2,3]

b = [6,7,8,9,10]

the desired result should be (if the next element in the cycle was 2):

[[6,2], [7,3], [8,1],[9,2],[10,3]]

So I, intuitively tried this:

2.0.0-p195 :469 > b = [6,7,8,9,10]
=> [6, 7, 8, 9, 10]
2.0.0-p195 :471 > a = [1,2,3].cycle
=> #<Enumerator: [1, 2, 3]:cycle>
2.0.0-p195 :472 > a.next
=> 1
2.0.0-p195 :473 > a.peek
=> 2
2.0.0-p195 :474 > b.zip(a)
=> [[6, 1], [7, 2], [8, 3], [9, 1], [10, 2]]

So, it seems that the enumerator a is starting from the beginning,
instead of from the "next". If I read correctly the source code (MRI)
it seems that zip uses take, and testing this shows that it always
starts from the beginning:

2.0.0-p195 :475 > a.take 2
=> [1, 2]
2.0.0-p195 :476 > a.take 2
=> [1, 2]
2.0.0-p195 :477 > a.take 2
=> [1, 2]

Is it wrong to expect the enumerator to behave as a cycle starting
from its current position for operations like zip and take?

Jesus.

This is not an answer for your question. But you can solve your original
problem like this:

a = [1,2,3].rotate(1).cycle

=> #<Enumerator: ...>

b = [6,7,8,9,10]

=> [6, 7, 8, 9, 10]

b.zip(a)

=> [[6, 2], [7, 3], [8, 1], [9, 2], [10, 3]]

Greets!

···

2014-06-26 13:21 GMT-03:00 Jesús Gabriel y Galán <jgabrielygalan@gmail.com>:

Hi,

I was surprised today with a behaviour, maybe my expectation was wrong
and someone can explain why it works this way:

I have an array of elements, and I want to cycle through it assigning
each element to another list of elements. But the cycle can be already
in the middle. For example:

a = [1,2,3]

b = [6,7,8,9,10]

the desired result should be (if the next element in the cycle was 2):

[[6,2], [7,3], [8,1],[9,2],[10,3]]

So I, intuitively tried this:

2.0.0-p195 :469 > b = [6,7,8,9,10]
=> [6, 7, 8, 9, 10]
2.0.0-p195 :471 > a = [1,2,3].cycle
=> #<Enumerator: [1, 2, 3]:cycle>
2.0.0-p195 :472 > a.next
=> 1
2.0.0-p195 :473 > a.peek
=> 2
2.0.0-p195 :474 > b.zip(a)
=> [[6, 1], [7, 2], [8, 3], [9, 1], [10, 2]]

So, it seems that the enumerator a is starting from the beginning,
instead of from the "next". If I read correctly the source code (MRI)
it seems that zip uses take, and testing this shows that it always
starts from the beginning:

2.0.0-p195 :475 > a.take 2
=> [1, 2]
2.0.0-p195 :476 > a.take 2
=> [1, 2]
2.0.0-p195 :477 > a.take 2
=> [1, 2]

Is it wrong to expect the enumerator to behave as a cycle starting
from its current position for operations like zip and take?

Jesus.

--
Juanjo Conti
blog: http://www.juanjoconti.com.ar

For the record: I asked in ruby-core and the answer is that it is a
design decision:

Marc-Andre Lafortune said:

"Your confusion is understandable, but this is not how the enumerators
were designed. At the end of the doc for `next` you'll find:

    Note that enumeration sequence by next does not affect other
non-external enumeration methods, unless the underlying iteration
methods itself has side-effect, e.g. IO#each_line

I think this design choice is in part because using `next` is much
slower then `each`."

So I guess that's it.

Jesus.

···

On Thu, Jun 26, 2014 at 6:34 PM, Carlos Agarie <carlos.agarie@gmail.com> wrote:

This is surprising behavior to me. In case no one finds a good explanation
in this mailing list, I think you should send a ticket to ruby-core asking
why it works this way (and fixing it in case there isn't a good reason to).