Improve perf of the interactive window UI

See info here: https://reactjs.org/docs/optimizing-performance.html

Run something like this:

#%%
import tensorflow as tf
mnist = tf.keras.datasets.mnist

(x_train, y_train),(x_test, y_test) = mnist.load_data()
x_train, x_test = x_train / 255.0, x_test / 255.0

model = tf.keras.models.Sequential([
  tf.keras.layers.Flatten(input_shape=(28, 28)),
  tf.keras.layers.Dense(512, activation=tf.nn.relu),
  tf.keras.layers.Dropout(0.2),
  tf.keras.layers.Dense(10, activation=tf.nn.softmax)
])
model.compile(optimizer='adam',
              loss='sparse_categorical_crossentropy',
              metrics=['accuracy'])

model.fit(x_train, y_train, epochs=1000)
model.evaluate(x_test, y_test)

And eventually we slow down so much it’s unusable.

[janosh]

This probably has a separate cause unrelated to React but it still strongly impacts interactive window performance: I’ve noticed that on sessions which create very large variables (like learning histories/traces of ML models) executing even simple cells in the presence of such variables becomes unusably slow just as described here.

Basically, if there is an extremely large variable in the notebook, every single cell will take a very long time to execute (> 5 seconds), no matter how simple the cell contents are. This is because the varInspector function runs the var_dic_list after every single cell execution, and var_dic_list is slow.

Not sure if VSCode Python can do anything about that but I thought I’d mention it.

[xgdgsc]

Thanks. The insider build works pretty well.