DBTreeList - A node created using dataset methods is inserted at an incorrect position if the tree list's OptionsData.SmartLoad property is set to True

Hello,
I am afraid that we cannot offer a more appropriate solution for this scenario. This behavior is caused by the limitation of the SmartLoad mode, and we cannot change it at the level of our controls without losing benefits of this mode. Let me explain how our SmartLoad feature works.
If the SmartLoad feature is enabled, the tree list is first populated only with root-level nodes. You can perform a small test to check this behavior. Just handle the form's OnShow event in your demo project and get the number of loaded nodes:

Delphi
Caption := IntToStr(DBTreeList1.AbsoluteCount);  

You will notice that if the SmartLoad property is enabled, this value is 2 (only root nodes). If the SmartLoad property is disabled, this value is 6 (all DataSet records are loaded). This allows us to improve performance and reduce memory usage. The algorithm of loading records for the root level looks like this:

Delphi
DataSet.Locate(ParentIDField, RootValue);  
while not DataSet.Eof and (ParentIDField.Value = RootValue) do  
begin  
  AddRootNode;  
  DataSet.Next;  
end;  

That is why it is necessary to have the underlying DataSet sorted by the ParentID field - the algorithm stops loading records when ParentIDField.Value is not equal to RootValue, and all "RootValue" records that do not reside sequentially will not be loaded.
 If a particular node is being expanded, the tree list loads its child nodes:

Delphi
DataSet.Locate((ParentIDField, ExpandedNode.ID);  
while not DataSet.Eof and ParentIDField.Value = ExpandedNode.ID do  
begin  
  ExpandedNode.AddChild;  
  DataSet.Next;  
end;  

As you can see in this code, we use the DataSet.Locate method to locate the first DataSet's record where the ParentID field value refers to the expanding node. Then child records are loaded sequentially.
Now let me explain what happens when you use the following code:

Delphi
M1.Append;  
M1.FieldByName('ParentId').AsString:='1';  

To be synchronized with the DataSet, TreeList focuses the newly created node. As this node is not the root node (ParentId is 1, but not 0), its parent node needs to be expanded to show this new node. As a result, the algorithm of loading child records I mentioned above is performed, and the current record of the underlying DataSet is changed (because of the Locate and Next methods). As a small test, add the root node as follows:

Delphi
M1.Append;  
M1.FieldByName('ParentId').AsString:='0';  

And in this case you will achieve the expected result. It is not necessary to expand any node as all root nodes are already loaded and do not need to be reloaded.
I am afraid that this behavior is by design and cannot be changed with the current structure of our ExpressQuantumTreeList. It seems that the SmartLoad mode is not suitable for the behavior you wish to implement.

Hello Paulo,
thank you for your explanation. I understead now how smartload works. My biggest problem at this moment is error : List index out of bounds, i have succeeded to create sample to demonstrate this. I am using 13.2 version.
Click on the test button, sometimes is necessary to click a few times.

Hello,
I have reviewed your project and see that you set the ID and ParentID fields for certain nodes to random values. I do not understand when this may be necessary in a real-world application. In this case, there is a chance that two nodes will have the same key value (the 'Code' field in your demo), but the key value should be unique. That is why the TreeList may not find a corresponding node when iterating through its node nodes.
In addition, I should say that it is not quite correct to use the BeginUpdate/EndUpdate block inside the loop body. It this case, the TreeList is notified about changes in each loop iteration. I suggest you use the following code instead to increase the performance:

Delphi
procedure TForm1.Button2Click(Sender: TObject);  
var  
i,j: integer;  
begin  
  DBTreeList1.BeginUpdate;  
  M1.DisableControls;  

  for j:=0 to 100 do  
  begin  
    M1.Append;  
//..  
    M1.Post;  
  end;  

  M1.EnableControls;  
  DBTreeList1.EndUpdate;  
end;  

This was just example, i using query (which is sorted at Parentid and indexed also) and the problem still exists. The second I can't put all inserts between DBTreeList1.BeginUpdate… DBTreeList1.Endupdat because they are separately activated in my application.
I need solution which will not produce nasty error if smart load is enabled. Alsi I did't experience this problem in the versions <1.8 and this application exists for few years.

Hello,
there is also another problem with smartload. In the example click on the "insert" button. When value of parent field is defined in the code, then dataset "springs" on the wrong record.
I really need fix for those two problems.

We are working on your request, but it can take us some time to examine it. I will get back to you once we have any results or need additional information. Thank you for your patience.