That's one viewpoint, and probably the more correct one. For example, the Python interpreter runs on a huge variety of machines but the machine you target with your code (platform-specific modules notwithstanding) is the Python "virtual" machine so you can run the same code on many disparate platforms.
The person that did all the heavy lifting there was Guido himself or, more generally, the person implementing the interpreter on each platform.
The other viewpoint is that there may be subtle variations in the platforms which will require you to adjust. Let's say the interpreter runs on two platforms, one with a 800x600 screen, the other with a 1920x1024 screen.
So, even though there may be a consistent plot
command, the differing resolutions may require some intelligence on your part to adjust to the specific machine.
In that case, you may well want to abstract the screen to a specific resolution so that your virtual screen is 4096x4096 and have that scaled to the actual resolution when plotting.
For example:
# Virtual plot, x and y are both 0-4095
def plot(x,y):
# Get physical screen dimensions.
actualWidth = getSysData ("screenwidth")
actualHeight = getSysData ("screenheight")
# Convert virtual co-ordinates to physical ones.
x = x * actualWidth / 4066
y = y * actualHeight / 4066
# Plot the physical ones.
sysplot (x, y)