Why do the trainable variables disappear when eager execution is enabled?

Question

If I run the following code (tensorflow 1.15) I can get a list of the trainable variables in two different ways.

from tensorflow.keras.layers import AveragePooling2D, Conv2D, Dense, Flatten, Input
from tensorflow.keras.models import Model
import tensorflow as tf

x_in = Input((32, 32, 1))
x = Conv2D(filters=6, kernel_size=(5, 5), activation='relu', input_shape=(32,32,1))(x_in)
x = AveragePooling2D(pool_size=(2, 2))(x)
x = Flatten()(x)
x = Dense(units=120, activation='relu')(x)
x = Dense(units=10, activation='softmax')(x)

m = Model(inputs=x_in, outputs=x)

v1 = m.trainable_variables
v2 = tf.compat.v1.trainable_variables()

Both v1 and v2 have the same value.

If I add in a call to tf.compat.v1.enable_eager_execution() before creating my model, v2 becomes empty; tf.compat.v1.trainable_variables() returns an empty list.

Why is this?