On 06/14/2013 12:55 PM, Paul Sandoz wrote:
On Jun 14, 2013, at 12:12 PM, Remi Forax <fo...@univ-mlv.fr> wrote:
The following does not throw CME:

             List<Integer> l = new ArrayList<>(Arrays.asList(1, 2));
             for (Integer i : l) {
                 l.remove(1);  // 2 is never encountered
             }

Where as the following does:

             List<Integer> l = new ArrayList<>(Arrays.asList(1, 2, 3));
             for (Integer i : l) {
                 l.remove(1);
             }

Because the hasNext implementation does not check for modification. It's sad 
this also occurs for the default implementation of Iterable.forEach :-(

This behaviour sucks.
devil advocate: why exactly, the iteration is finished when you remove the 
element ?
The latter because a CME is thrown; the former because hasNext returns false.

The above is an example of how a bug can be hidden depending on the state (# 
elements) of the collection.


It would be a shame for overriding forEach/forEachRemaining implementations to 
conform to such behaviour when they can implement stronger/consistent failure 
guarantees.
While I could agree with you in theory, in practice I have seen several times 
codes that rely on this behaviour,
usually there is a bunch of method calls between the for loop and the 
list.remove() so this is not something that can be easily fixed.
A bug none the less, yes?

In the codes I was referring to, there was always a way to know that the remove was done at the end by example by knowing that the last element was a special sentinel or by using a counter.
So is the following program bugged ?

List<Integer> l = new ArrayList<>(Arrays.asList(1, 2, null));
for (Integer i : l) {
  if (i == null) {
    l.set(l.size() - 1, 3);  // change the last value to 3
  }
}


And because I think it's more important that users should be able to use either 
for(:) or forEach without thinking too much,
because otherwise nodoby will "modernize" their code, we have no choice but 
stick to the iterator behaviour for forEach
i.e. no modCount check at the end.

So you argument is based on the the premise that existing buggy code should 
continue work?

buggy code or not buggy code, that is the question.


Paul.

RĂ©mi

Reply via email to