Multiple rerenders when using ipywidgets.Output.append_stdout

[danr]

When you put a widget in a ipywidgets.Output and using append_stdout on the output the widget will be rerendered once per append.

cell 1:

import anywidget
import traitlets

import ipywidgets as iw 
import IPython.display as ipd

class TinyWidget(anywidget.AnyWidget):
    _esm = """
    let num_created = 0
    function render({ model, el }) {
      const my_creation_id = ++num_created;
      const div = document.createElement("div");
      div.innerHTML = `my_creation_id: ${my_creation_id}`;      
      el.appendChild(div);
    }
    export default { render };
    """
    
out = iw.Output()
with out:
    ipd.display(TinyWidget())
out

cell 2:

for i in range(5):
    out.append_stdout(f'{i} ')

Expected output:

my_creation_id: 1
0 1 2 3 4 

Observed output:

my_creation_id: 6
0 1 2 3 4 

Notes:
This only happens when the appends are in a separate cell.

I noticed this and while this is mostly a bug upstreams I wanted to document it here as an issue. There are many open issues with ipywidgets.Output. I don't know exactly which one is relevant in this case but this comment is a good starting point for these issues: jupyter-widgets/ipywidgets#3261 (comment)

In line with that comment, the expected output happens if you use the context manager instead of the append_stdout method, like this:

for i in range(5):
    with out:
        print(f'{i} ', end='')

[manzt]

Hi there, thanks for opening the issue! Indeed, this is an implementation detail wrt to how ipywidgets.Output works. In this case, I'd recommend using an ipywidgets.VBox to avoid re-returns (for the custom Jupyter Widget):

ipywidgets.VBox([TinyWidget()], out])

[asaboor-gh]

I am not sure if anywidget reloads javascript with each render or just runs the render function. If it is just render function, and your num_created is outside render, it is logical to have observed output. Let me explain:

ipywidgets.Output has an outputs attribute that get updated on each time a new display object is created. That attribute is a tuple which is assigned again and again each time an append_<func> is called. Since you are updating 5 times but it was also shown in start, so it becomes 6th time when you changed the outputs attribute. It's infact not appending new objects, it's re-assigning the outputs with all object appended so far.