RollingBeta x and y are mixed up

[Javdat]

From function definition:

x | A vector, matrix, list or other object coercible to a matrix; the independent variable
y | A vector, matrix, list or other object coercible to a vector; the dependent variable

But, the results are opposite:

x = rnorm(100)
y = x * 2
RollingBeta(x = x, y = y, window = 50)

expected result 2 (as we have defined y = x*2), actual result: 0.5.

Hence definition of x and y is opposite.

This gives correct results:

RollingBeta(x = y, y = x, window = 50)

I wondered why tests pass. This is the result.
From unit tests:

RollingWindow/tests/test.R

Line 65 in 6c3adda

Beta = function(lhs, rhs) cov(lhs, rhs) / var(rhs)

Tests call it as (x, y) hence lhs = x

RollingWindow/tests/test.R

Line 249 in 6c3adda

run_tests_xy(x, y, window, funcs_xy, trace = TRACE)

The correct formula for simple linear regression is: cov(x,y)/var(x)
Hence (note var(lhs)):

Beta = function(lhs, rhs) cov(lhs, rhs) / var(lhs)

The main culprit:

RollingWindow/src/RollingWindow.cpp

Line 552 in 6c3adda

case BETA : rollxy[i] = cov / var_y; break;

Should be cov/var_x