3 ms·
And it does not cause an error, despite being written back when bytes mattered. Hats off, anonymous Nintendo programmer, you are the modern equivalent of medie
by patio11 16y ago
And it does not cause an error, despite being written back when bytes mattered. Hats off, anonymous Nintendo programmer, you are the modern equivalent of medieval artists who painted parts in cathedrals which would be occluded by planned construction "because God would see if they were unpainted."
- unwind 16y agoToshihiko Nakago, Kazuaki Morita and Yasunari Nishida were the main programmers according to <http://www.gamefaqs.com/nes/525243-super-mario-bros> http://www.gamefaqs.com/nes/525243-super-mario-bros>.
- pilif 16y agoI'm not sure whether that kind of being error-proof was actually intended. The point is that the data structure of the level was so simple that basically any memory location could be interpreted as level data. There's another glitch with the warp pipes that you can enter before the location they point to has been completely loaded at which time you basically warp to a more or less random memory address leading to this: http://tasvideos.org/1365M.html http://tasvideos.org/1365M.html In that case though, if you are not really careful, you can actually quite easily crash an emulator which highly likely also would cause the NES to crash.
- bphogan 16y agoDunno about that. I have played the "Minus World" since 1990 when I found out how to do it from my friend. I try it on tons of emulators too, and they all seem to handle it pretty well. I don't recall crashing an emulator with it, and I've never crashed the NES. (Of course it does just get stuck in a big endless loop since the level never advances when you get to the end, but it's a cool trick to do nonetheless).
- Luc 16y agoThere is an awesome disassembly of the code available: http://www.romhacking.net/docs/344/ http://www.romhacking.net/docs/344/ . Almost every line is commented, so it's possible to follow along without knowing too much 6502. I took a very quick look at it. If you search for 'L_GroundArea1', you'll find the level data for area 3-3. Notice the end-of-data marker is $fd. I then searched for this value, and found the 'ProcessAreaData' subroutine compares against it. By the look of it, this routine decodes new level tiles, except if $fd is found, in which case it just leaves the current tile in the buffer.