Comments on: Odd Rubyism

By: James Robey

James Robey — Thu, 04 Feb 2010 14:57:55 +0000

Well, an item is being deleted from the array while the array itself is being used for the loop.

Basically, it’s behaving normally, the array is shrinking, while the internal count is staying the same.

[0,1,2], count is 0 item = 0

delete 0, then increment count at end of loop

[1,2], count 1 item = 2

Hashes are unordered, so perhaps it doesn’t use an internal counter at all, but a pointer instead. Maybe that’s why the behavior is different.

By: John Stehle

John Stehle — Wed, 11 Mar 2009 20:11:43 +0000

Yes! that is not what one would expect. I ran across this behavior and did not know if it was as designed or not. What I do is nil the elements I want to delete then do a compact! on the array.

By: Bill Kayser

Bill Kayser — Sun, 04 Jan 2009 18:53:51 +0000

Java iterator behavior is well-defined and predictable. There are different iterator types that support different features. Generally if you delete or add an underlying element while iterating the iterator will throw a concurrent modification exception.

Other iterator types support things like going forward and backward, inserting at the current index, and deleting the current item–without throwing the exception. That allows you to avoid the copy on read pattern to avoid the exception.

By: José Valim

José Valim — Fri, 02 Jan 2009 18:30:11 +0000

Very odd inded!
Have you checked Ruby 1.9 if we have the same behavior? =)