z26 versus Stella 2.8.4:
The purple slanty-line on the right is drawn with missile1 (the one on the left is drawn with player1), and it should have been drawn 4 color clocks to the right, so it looks like 2.8.4 had an occasional problem positioning missile1-- an examination of the disassembly for the section of the kernel where that part of the screen is being drawn might reveal more clues about exactly what was being done there that may have been a bit different than what was being done in other places. In any case, that particular glitch has been fixed in the 3.0_alpha1 build:
z26 versus Stella 3.0_alpha1:
The other two glitches are still there. The yellow lines on the left are drawn with missile0, and everything looks kosher except for where the line starts to slant inward-- the line's position is okay, but it should be twice as wide (i.e., an HMOVE and a NUSIZ0 change occur on the same scan line).
The purple flippers are drawn with player1, but I'm not certain what the glitch is. It looks like player1 is being drawn with two extra copies, and the graphics for the second extra copy aren't being cleared soon enough, such that the two leftmost bits from the first extra copy are still there. Except, I can't figure out why the programmer didn't just use one extra copy, instead of two extra copies, in which case there'd be no need to clear the graphics before another copy was displayed. Again, a look at a disassembly of the program might reveal more clues about what's being done in that section of the kernel.
Michael
















