The source code on github for Chapter 15 may have some inconsistency

I was following the examples in chapter 15 and found that there may be some inconsistency in predicting future values.
In cell 24, the following code is used in making prediction:
for step_ahead in range(10):
y_pred_one = model.predict(X[:, step_ahead:])[:, np.newaxis, :]
X = np.concatenate([X, y_pred_one], axis=1)

I believe that we are feeding the model with an input vector of 50 features, as we concatenate y_pred_one to X, we slide the feature input using X[:, step_ahead:] and keep its size at 50.

But in cell 28, which uses the same model, the source code is:
for step_ahead in range(10):
y_pred_one = model.predict(X)[:, np.newaxis, :]
X = np.concatenate([X, y_pred_one], axis=1)

In this snippet, the size of input X is increasing as we concatenate y_pred_one to it. I checked the outputs (with sliding the input set and w/o sliding it), they are different.
My question is which one is correct? And why?