To clarify, you're doing the wrong calculation. Technically speed is the derivative of position with the respect to time, but over small enough changes in position we can approximate it as the change in position - distance - over the change in time. In this case assume
x is the position of light. Then
c=dtdx≈ΔtΔx.In your notation, you can write this as
c=td.
Now consider two different observers in difference reference frames. The first observer, Alice, measures the speed of light to be
c and the distance and time to be
d1 and
t1 respectively. The second observer, Bob, also measures the speed of light to be
c (by the postulate of SR, as confirmed experimentally). He measures the distance and time to be
d2 and
t2. Where is the contradiction here?
t1d1=c=t2d2.
We only need
t1=t2 if
d1=d2, but this is manifestly not the case for different reference frames! There are infinitely many variations that leave the speed of light invariant. For example, if
d2=2d1 and
t2=2t1,
t2d2=2t12d1=t1d1 and there is no problem.