In the diffraction grating experiment, it is assumed that the light enters the slit as a 'plane wave'. This means that for a uniformly radiating light source (a light bulb, for example) the ever expanding circle of light represented by the leading wave front enters the slit in a ~straight line. You can see this set-up on Dr. Rod Nave's Georgia State University hyperphysics site.
Think about the other extreme: If the light source was parallel to the slit, then no light would enter the slit at all!
The placement of the light source also has an effect on the equations that can be used.
For example, the maxima equation: `dsintheta=m lambda`
(where the theta is the angle of the maxima) assumes that the light source is normal to the slit (i.e. perpendicular to).
If the light source is at some angle to the diffraction grating, the equation gets more complicated: `d (sin theta_(i) + sin theta_(m)) = m lambda`