Differences

This shows you the differences between two versions of the page.

--- bunker [2016/05/25 16:56]
ex_writer [Code Discussion]
+++ bunker [2016/05/26 20:06]
ex_writer [Result]
@@ Line 1: / Line 1: @@
-====== Space Invaders Bunker ======
+====== Space Invaders Bunker Snippet ======
 The following (edited) code snippet is from Richard Degler's [[http://www.ballyalley.com/ml/ml_source/Space%20Invaders%20(197x)(Bally%20Mfg.%20Corp.)(proto)%5BDisassembly%5D.zip|disassembly]] [.zip] of the Bally Arcade Space Invaders prototype. The LOOP section demonstrates writing a single pattern (the bunker graphic) to the screen multiple times using the [[screen handler#writp|WRITP]] UPI routine. The BUNKER section shows the referenced pattern block.
@@ Line 55: / Line 55: @@
 ===== Code Discussion =====
-The first line loads register B with 4 for use as the loop counter. We want to draw the same bunker image four times along the bottom of the screen. Each will have identical y-coordinates, but increasing x-coordinates (as we draw left to right).
+The first line loads register B with 4 for use as the loop counter. We want to draw the same bunker image four times along the bottom of the screen. Each will have identical y-coordinates, but increasing x-coordinates (as we draw left to right), so the second line sets the initial value of register E to 10. As we'll see below, E will serve as storage for the x-coordinate when we call the pattern drawing routine.
-[[screen handler#writp|WRITP]] is squarely in the middle of the [[screen write routines|screen writing routine]] hierarchy, meaning that it some preprocessing to execute properly.
+The third line marks the LOOP label, which is where the program will return again and again until B reaches 0. This also marks the point where we begin 'pre-loading' our registers with the arguments for our [[screen handler#writp|WRITP]] call. WRITP is squarely in the middle of the [[screen write routines|screen writing routine]] hierarchy, meaning that it requires some housecleaning before we call it.
-[pattern block requirements]
+Line 3 loads the address of the BUNKER pattern block in HL. WRITP assumes that the pattern size is stored //with the pattern block//, so the first two bytes at label BUNKER ($05, $0D) define the bunker graphic's width (5) and heigh (13) in bytes.
+Line 4 'loads' the accumulator with the [[Magic Register]] settings for our pattern write. XOR A is a more efficient way to perform LD A,0 (though the former does change flags). In other words, we're sending a 0 to the Magic Register, indicating that we want to simply 'PLOP' the pattern onto the screen with no transformations.
+Line 5 load the y-coordinate that we'll use for for every bunker (65) in the D register, as required by WRITP.
+Lines 6 and 7 push the contents of DE and BC onto the stack to preserve their contents during the WRITP call.
+Line 8 finally invokes WRITP now that all the necessary pattern information is in place: the pattern's size, contents, position, and (lack of) transformations.
+Lines 10 and 11 pop the contents of BC and DE off the stack.
+Line 12 loads the value of the current x-coordinate from E to A, the following line adds 40 to that value, then line 14 puts the sum back into E. As a result, on the next iteration through the loop, the bunker's x-coordinate will be 40 pixels to the right of its previous location.
+The final line of the LOOP checks to see if B is zero. If not, the program proceeds to the beginning of LOOP; if so, program execution proceeds.
 [why choose WRITP vs. others]
-<code z80>XOR  A</code> is a more efficient way to perform <code z80>LD  A,0</code>, though the former does change flags.
+===== Output =====
+{bunker.png}