Deleting a node in foreach loop - c#

Look at this csharp code, and see if you can tell why I need to exit the loop after I found and deleted an item from the list. The idea is to go through a node's list of neighbors, and see if a Node n exists there, then delete that neighbor:
internal void RemoveDirected(Node n)
{
foreach (EdgeToNeighbor etn in this.Neighbors)
{
if (etn.Neighbor.Key == n.Key)
{
RemoveDirected(etn);
break;
}
}
}
internal void RemoveDirected(EdgeToNeighbor e)
{
Neighbors.Remove(e);
}
.
.
.
// Removes EdgeToNeighbor instance from AdjacencyList
protected internal virtual void Remove(EdgeToNeighbor e)
{
base.InnerList.Remove(e);
}
Notice how I have a "break" after the RemoveDirected call in the first method.
I've found that if I didn't exit after the RemoveDirected
it would go on forever in the foreach loop. I suppose it must have
something to do with the way foreach works. If you modify the list
that foreach is working on, it gets confuse and loops forever.
Have you seen this type of thing, and what are other options to use rather than using break?
Of course, I could place the node that I've found in a local variable, then break from the loop, and
delete it outside of the loop. But I was thinking, may be there are better ways to do this
in csharp.

When you iterate a .NET collection using its iterator, you must not modify that collection. If you do, you are asking for trouble.
You should defer the deletion instead of deleting right in the foreach loop. For example, you can collect everything you need to delete in a list, and then delete it outside of foreach.
var toDelete = this.Neighbors.Where(etn => etn.Neighbor.Key == n.Key).ToList();
foreach (var e in toDelete) {
Neighbors.Remove(e);
}

You definitely don't want to use an iterator if you are removing items. One option is to change it to downward counting loop:
for (int nI = this.Neighbors.Count; nI >= 0; nI--)
{
var etn = this.Neighbors[nI];
if (etn.Neighbor.Key == n.Key)
{
RemoveDirected(n);
}
}
You could also collect the keys or items to be deleted in a collection within the loop and then perform the deletion once you have completed your iteration.
However, if you are only removing a single item and the items in the collection are keyed somehow, then you shouldn't actually need any loop. Just test the existence of the key or item in the collection and, if it is there, just remove it.

You could get the number of nodes and then check and delete from highest to lowest in a for loop that way you avoid looking for an item that no longer exists.

Related

Threading Dictionary<string,int> and locks

I have a dictionary that is being used in two threads, one is sending out UDP packets, one is receiving them. Both are keeping a common collection to see count outgoing and returning packets and hopefully keeping them at 0 difference :)
Now I'm iterating through the dictionary to update values and after iteration it errors. I do have a lock object in place, how could I solve this?
First Thread:
lock (retryListLock)
{
// loop through all known devices in the device list to build a counter if the device still lives
foreach(string key in retryList.Keys)
{
retryList[key] += 1;
if (retryList[key] > Retries)
{
DiscoveredDevice device = Devices.Find(d => d.SerialNo == key);
if (device != null)
{
OnDeviceRemoved(device);
Devices.Remove(device);
retryList.Remove(key);
}
}
}
}
Second Thread:
lock (retryListLock)
{
if (retryList.ContainsKey(frame.SerialNo))
retryList[frame.SerialNo] = 0;
else
retryList.Add(frame.SerialNo, 0);
}
I'm only getting the error after the first thread adds +1 to the value of that item, in the second iteration it errors out:
the collection has changed. enumeration operation may not execute (translated from Dutch)
How can I solve this? Obviously the Dictionary is the easiest to use for me in this case.
The problem is that you cannot change the dictionary in an iterator/foreach
foreach(string key in retryList.Keys)
{
retryList[key] += 1; // <-- The error happens here ! Do not alter the Dictionary during an iteration
if (retryList[key] > Retries)
{
DiscoveredDevice device = Devices.Find(d => d.SerialNo == key);
if (device != null)
{
OnDeviceRemoved(device);
Devices.Remove(device);
retryList.Remove(key); // <-- The error could also happen here ! Do not alter the Dictionary during an iteration
}
}
}
I found this question on stackoverflow which might help you
How to iterate through Dictionary and change values?
Here we find a statement from MSDN
The foreach statement is a wrapper around the enumerator, which allows
only reading from the collection, not writing to it.
With thanks to Bongo, taking the keys into a second list for iteration solved it:
List<string> Keys = new List<string>(retryList.Keys);
foreach(string key in Keys)
The error has nothing to do with the locks and multithreading.
Your enumerator (what foreach are using) is invalidated when you modify the very same data structure (the dictionary) on what the enumerator enumerates in the loop.
Solution:
You must first loop say with foreach, and remember practically in a list what you should remove. Then in a separate loop on the remembered keys, remove they from the dictionary.
This might be useful to you. Concurrent Dictionary
It's thread safe
Also change the for each to a reverse for loop
Something like this.
for (int x = max; x>=1; x--)
{
}

How add or remove object while iterating Collection in C#

I am trying to remove object while I am iterating through Collection. But I am getting exception. How can I achieve this?
Here is my code :
foreach (var gem in gems)
{
gem.Value.Update(gameTime);
if (gem.Value.BoundingCircle.Intersects(Player.BoundingRectangle))
{
gems.Remove(gem.Key); // I can't do this here, then How can I do?
OnGemCollected(gem.Value, Player);
}
}
foreach is designed for iterating over a collection without modifing it.
To remove items from a collection while iterating over it use a for loop from the end to the start of it.
for(int i = gems.Count - 1; i >=0 ; i--)
{
gems[i].Value.Update(gameTime);
if (gems[i].Value.BoundingCircle.Intersects(Player.BoundingRectangle))
{
Gem gem = gems[i];
gems.RemoveAt(i); // Assuming it's a List<Gem>
OnGemCollected(gem.Value, Player);
}
}
If it's a dictionary<string, Gem> for example, you could iterate like this:
foreach(string s in gems.Keys.ToList())
{
if(gems[s].BoundingCircle.Intersects(Player.BoundingRectangle))
{
gems.Remove(s);
}
}
The easiest way is to do what #IV4 suggested:
foreach (var gem in gems.ToList())
The ToList() will convert the Dictionary to a list of KeyValuePair, so it will work fine.
The only time you wouldn't want to do it that way is if you have a big dictionary from which you are only removing relatively few items and you want to reduce memory use.
Only in that case would you want to use one of the following approaches:
Make a list of the keys as you find them, then have a separate loop to remove the items:
List<KeyType> keysToRemove = new List<KeyType>();
foreach (var gem in gems)
{
gem.Value.Update(gameTime);
if (gem.Value.BoundingCircle.Intersects(Player.BoundingRectangle))
{
OnGemCollected(gem.Value, Player);
keysToRemove.Add(gem.Key);
}
}
foreach (var key in keysToRemove)
gems.Remove(key);
(Where KeyType is the type of key you're using. Substitute the correct type!)
Alternatively, if it is important that the gem is removed before calling OnGemCollected(), then (with key type TKey and value type TValue) do it like this:
var itemsToRemove = new List<KeyValuePair<TKey, TValue>>();
foreach (var gem in gems)
{
gem.Value.Update(gameTime);
if (gem.Value.BoundingCircle.Intersects(Player.BoundingRectangle))
itemsToRemove.Add(gem);
}
foreach (var item in itemsToRemove)
{
gems.Remove(item.Key);
OnGemCollected(item.Value, Player);
}
As the other answers say, a foreach is designed purely for iterating over a collection without modifying it as per the documenation:
The foreach statement is used to iterate through the collection to get
the desired information, but should not be used to change the contents
of the collection to avoid unpredictable side effects.
in order to do this you would need to use a for loop (storing the items of the collection you need to remove) and remove them from the collection afterwards.
However if you are using a List<T> you could do this:
lines.RemoveAll(line => line.FullfilsCertainConditions());
After going through all the answers, and being equally good. I faced a challenge where I had to modify a List and what I ended up doing worked quite well for me. So just in case anyone finds it useful. Can someone provide me feedback on how efficient it might be.
Action removeFromList;
foreach(var value in listOfValues){
if(whatever condition to remove is){
removeFromList+=()=>listOfValues.remove(value);
}
}
removeFromList?.Invoke();
removeFromList = null;
You should use the for loop instead of the foreach loop. Please refer here
Collections support foreach statement using Enumarator. Enumerators can be used to read the data in the collection, but they cannot be used to modify the underlying collection. If changes are made to the collection, such as adding, modifying, or deleting elements, the enumerator is irrecoverably invalidated and the next call to MoveNext or Reset throws an InvalidOperationException.
Use for loop for collection modifying.

Modifying list inside foreach loop

I have a construction similar to this (but a lot more complicated):
var list = new List<string>();
// .. populate list ..
foreach(var item in list)
{
DoFunction(list);
}
public void DoFunction(List<string> list)
{
if(someCondition == true)
{
// .. modify list in here ..
}
}
Now, I understand that its not possible to edit the collection you're foreaching through, but how do you jump out of the loop gracefully if you do have to edit the list (without a try catch statement)? Is there some way to tell if the list has been edited? Can you edit the list and quickly break; before it notices?
Yes, you could break, if that's what you really want. An exception won't be thrown until the for loop tries to grab the next item from the list.
But I've found it's easiest just to create and iterate across a copy of the list so you don't have to worry about it.
foreach(var item in list.ToList())
The added performance overhead of an extra, untouched list is generally negligible compared to the maintainability costs of more complex code.
Rather than use a foreach construct, a for loop would allow you to alter the list.
for (var x = 0; x < list.Count; x++) {
}
It's hard to offer useful advice without knowing what kinds of edits are being made. The pattern that I've found is has the most general-purpose value, though, to just construct a new list.
For example, if you need to look at each item and decide between removing it, leaving it as-is, or inserting items after it, you could use a pattern like this:
IEnumerable<string> butcherTheList(IEnumerable<string> input)
{
foreach (string current in input)
{
if(case1(current))
{
yield return current;
}
else if(case2(current))
{
yield return current;
yield return someFunc(current);
}
// default behavior is to yield nothing, effectively removing the item
}
}
List<string> newList = butcherTheList(input).ToList();

Strange behaviour between foreach and for loop

My windows phone 7 silverlight app before placing pushpins on a map layer removes any that are previously there.
I was doing this in a foreach loop as follows:
//Clear previous pins
try
{
foreach (UIElement p in PushPinLayer.Children)
{
if(p.GetType() == typeof(Pushpin))
{
PushPinLayer.Children.Remove(p);
}
}
}
catch (Exception ex)
{
MessageBox.Show(ex.Message);
//TODO: For some reason the foreach loop above causes an invalid Operation exception.
//Cathing the error here until I can work out why it is happening.
}
This code removes any pushpins as required but after the last loop would throw an exception "Invalid Operation" I re-wrote it as a for loop:
for (int i = 0; i < PushPinLayer.Children.Count; i++)
{
if (PushPinLayer.Children[i].GetType() == typeof(Pushpin))
{
PushPinLayer.Children.RemoveAt(i);
}
}
Which works fine, however I can't see why the foreach is throwing an error.
This is very normal,
You cannot remove items from a list which you still use in the foreach list.
better then removing the item would be to create a new list, and every time it is not a pushpin type, add the object to the new list.
This way the original list is not altered and you won't get an exception.
I find it strange that the for loop works, but if it does, it would mean that the way they are itterated is different. The for loop will be copied to another memory location and used for the for loop so that the original one, in which you remove items, is not used anymore by the for loop. The foreach loop will get parameters from the list, the you remove items, so the list and the parameters become concurrent.
Your foreach loop uses and Enumerator to iterate the objects in your collection. When you delete an object from the collection the Enumerator is no longer valid as it references object that no longer exist. This causes a InvalidOperationException
The best way to remove them is using a for loop and even better doing it in reverse.
for (int i = PushPinLayer.Children.Count - 1; i >= 0 ; i--)
{
if (PushPinLayer.Children[i].GetType() == typeof(Pushpin))
{
PushPinLayer.Children.RemoveAt(i);
}
}
This will ensure that as the items are removed your Index i does not exceed the number of items in your collection.
Since others have already answered your question, I will only comment on your usage of xxx.GetType() == typeof(Pushpin). You could check whether something is a pushpin by simply using the C# reserved keyword is. e.g.:
if (p is Pushpin) {...}
When you change the content of the collection, the enumerator used in the foreach loop becomes invalid. You can't change a collection while you enumerate it.
Here's a workaround:
List<UIElement> toRemove = new List<UIElement>();
foreach (UIElement p in PushPinLayer.Children)
{
if(p.GetType() == typeof(Pushpin))
{
toRemove.Add(p);
}
}
foreach(UIElement p in toRemove)
{
PushPinLayer.Children.Remove(p);
}
Alternatively, you could use the RemoveAll method, which takes a predicate as a parameter:
PushPinLayer.Children.RemoveAll(p => p is Pushpin);
Others have already given you the reason for the problem so I just thought I'd post a LINQ version of a work around
var toRemove = PushPinLayer.Children.OfType<Pushpin>().ToList();
// since toRemove is a separate collection, it's safe to do this now:
foreach (var child in toRemove)
PushPinLayer.Children.Remove(child)
You must not change the collection while foreaching through it.

Problems removing elements from a list when iterating through the list

I have a loop that iterates through elements in a list. I am required to remove elements from this list within the loop based on certain conditions. When I try to do this in C#, I get an exception. apparently, it is not allowed to remove elements from the list which is being iterated through. The problem was observed with a foreach loop. Is there any standard way to get around this problem?
Note : One solution I could think of is to create a copy of the list solely for iteration purpose and to remove elements from the original list within the loop. I am looking for a better way of dealing with this.
When using List<T> the ToArray() method helps in this scenario vastly:
List<MyClass> items = new List<MyClass>();
foreach (MyClass item in items.ToArray())
{
if (/* condition */) items.Remove(item);
}
The alternative is to use a for loop instead of a foreach, but then you have to decrement the index variable whenever you remove an element i.e.
List<MyClass> items = new List<MyClass>();
for (int i = 0; i < items.Count; i++)
{
if (/* condition */)
{
items.RemoveAt(i);
i--;
}
}
If your list is an actual List<T> then you can use the built-in RemoveAll method to delete items based on a predicate:
int numberOfItemsRemoved = yourList.RemoveAll(x => ShouldThisItemBeDeleted(x));
You could use LINQ to replace the initial list by a new list by filtering out items:
IEnumerable<Foo> initialList = FetchList();
initialList = initialList.Where(x => SomeFilteringConditionOnElement(x));
// Now initialList will be filtered according to the condition
// The filtered elements will be subject to garbage collection
This way you don't have to worry about loops.
You can use integer indexing to remove items:
List<int> xs = new List<int> { 1, 2, 3, 4 };
for (int i = 0; i < xs.Count; ++i)
{
// Remove even numbers.
if (xs[i] % 2 == 0)
{
xs.RemoveAt(i);
--i;
}
}
This can be weird to read and tough to maintain, though, especially if the logic in the loop gets any more complex.
Another trick is to loop through the list backwards.. removing an item won't affect any of the items you are going to encounter in the rest of the loop.
I'm not recommending this or anything else though. Everything you need this for can probably be done using LINQ statements to filter the list on your requirements.
You can iterate with foreach this way:
List<Customer> custList = Customer.Populate();
foreach (var cust in custList.ToList())
{
custList.Remove(cust);
}
Note: ToList on the list of variables, this iterates through the list created by the ToList but removes the items from the original list.
Hope this helps.
The recommended solution is to put all your elements you want to remove in a separate list and after the first loop, put a second loop where you iterate over the remove-list and remove those elements form the first list.
The reason you get an error is because you're using a foreach loop. If you think about how a foreach loop works this makes sense. The foreach loop calls the GetEnumerator method on the List. If you where to change the number of elements in the List, the Enumerator the foreach loop holds wouldn't have the correct number of elements. If you removed an element a null exception error would be thrown, and if you added an element the loop would miss an item.
If you like Linq and Lamda expressions I would recommend Darin Dimitrov solution, otherwise I would use the solution provided by Chris Schmich.

Categories