Loaded scripts identation broken once second program starts

[isidorn]

Refs: #54921

1. Start debugging a first program, loaded scripts shows up and all is good
2. Start debugging a second program, loaded scripts has broken identation (too much to the left) 🐛
3. Notice that if I expand / collapse some nodes the indentation gets fixed

[weinand]

that's a tree layout issue.

[isidorn]

@weinand None of the other trees are exibithing this behavior - and some do a very similar rendering to you, example Call Stack.
I suggest you look into this and check what is differnet between CallStack and Loaded Scripts before forwarding to Joao who does not fix tree bugs to my understanding.

[weinand]

The loaded scripts view is no longer extension code, so this is a native tree issue.
I'll investigate whether I can work around this.

[joaomoreno]

@weinand Interesting! Let me know if you come up with any insights.

[weinand]

@joaomoreno I've investigated this and found the following:
the layout gets broken if elements are reparented without changing their ID.

So assigning unique IDs to tree elements once, prevents that they can be reparented successfully later.
Changing my code to calculate IDs dynamically based on their parent chain fixed the problem.

In my case the the IDataSource of the initial tree was this (IDs in parenthesis):

Root (1)
   Source1 (3)
   Source2 (4)

the reparenting introduces a new intermediate "Session1" node (because another session 2 made this necessary):

Root (1)
   Session1 (2)
      Source11 (3)
      Source12 (4)
   Session2 (5)
      Source21 (6)
      Source22 (7)

The visual appearance with the broken layout) was this:

> Session1 (2)
Source11 (3)
Source12 (4)
> Session2 (5)
    Source21 (6)
    Source22 (7)

It looks like the indent of the elements Source11 and Source12 is not adjusted because they exist already in the tree.

Is this a bug or is there a constraint on tree element IDs?

[joaomoreno]

🤔

It could very well be an unintended constraint... I'm glad you found a way around it!

[weinand]

It would be helpful if you could add this constraint as a comment to the DataProvider interface.