Is it a bug to implement a transposed direct-form I (TDF-I) in Simulink?

On the Mathworks page (https://www.mathworks.com/help/signal/ref/dfilt.df1t.html), the description of the dfilt.df1t function shows the block diagram of the form:

but in the classical representation, the minus sign is placed before the denominator coefficients:

.

Thus, if in the implementation of Matworks with the initial states of the numerator and denominator sides equal to 5, the first output will be ():

then in the classical form under the same conditions the first output is

.

Wouldn't it be more correct to move the minus sign to the gain block?