Talk:999: Nine Hours, Nine Persons, Nine Doors

Things Worth Mentioning
The video for test3.fsb is just under 15 minutes. I don't know if it would be better to break it up into multiple videos or leave it as is. The other thing is that someone should translate the kanji and kana in the "Debug and Tests" section. I am curious to what the image and videos say. --Cuber456 23:33, 14 January 2013 (EST)

Codes Don't Work
I have followed the guide to the letter, but the codes do not work for me. I am using the DesMuMe 0.9.8 emulator and playing the Japanese version. I would appreciate it if someone helps me with this problem. 66sixx (talk) 23:54, 18 December 2013 (EST)
 * I can help but can you be a bit more specific? There are a bunch of codes listed on the main page and the notes page so I don't know what codes you tried and what codes you haven't tried. For me, the codes do work for the Japanese version BUT I am using DesMuMe 0.9.7, not 0.9.8. I just tried to use the codes in the latest version of DesMuME, which is 0.9.10, and they don't work. I'm guessing the codes will only work for that version of the emulator so if you want to try the codes out, run the game using DesMuMe 0.9.7. If it still doesn't work, you can say so but it should work. I should probably put a note on the main page and elsewhere about this since I didn't even know about this issue... and I created the codes.--Cuber456 (talk)
 * It appears DesMuME is not very well bug-tested. I have noticed other issues too. --Hiccup (talk) 04:03, 19 December 2013 (EST)
 * Well I guess I shouldn't be too surprised that it isn't well bug tested. It still sucks though. I was hoping that if I created these codes, they would be good for any version of the emulator. I guess I am naive. Unless there is an emulator that doesn't screw up cheat codes with every new version, like hell I'm going to recreate these codes. Its easy for one code but tedious as hell to recreate all of them.--Cuber456 (talk)
 * From the programmers:
 * "Between 0.9.7 and 0.9.8 the method used by the cheat system changed to one which applies the NDS unaligned access logic. I'm not sure there was a great reason for this."
 * So there's doubt that the codes would work on an actual DS anyway. Could someone confirm this? -Einstein95 (talk) 15:05, 19 December 2013 (EST)
 * I can't test them on an actual DS unfortunately. As a hypothetical, lets assume they don't work. Would the codes be removed if they didn't work on a real DS? I think it is kind of nice having at least some kind of method to access the unused stuff in the game. Is there any DS emulator that mimics the DS's memory and an action replay relatively accurately so that a code created in that emulator also works for the real hardware? What I am trying to say is that, if such an emulator exists, I wouldn't mind recreating the codes using that emulator since it would be kind of cool to be able to load this stuff on both the real hardware and emulator instead of just an emulator. --Cuber456 (talk)
 * Judging from what I've seen, the best bet would be DeSmuME, and making sure to keep up to date. You could always try no$gba, but no promises there. -Einstein95 (talk) 15:30, 19 December 2013 (EST)
 * Ok, I've investigated things a bit more and the problem as to why the Japanese version wasn't loading the unused data (and was crashing) is now obvious to me. With the newer versions of DesMuMe past 0.9.7, the action replay codes are aligned. For example, the code for Test2 on the main page (at the moment) is
 * 021E61C7 74736554 121E61CB 00000032
 * where the addresses in memory this code accesses is (supposed to be) 021E61C7 and 021E61CB. In the cheat code, 021E61C7 (essentially) says to start writing the value 74736554 starting at the address 021E61C7. In 0.9.7, it directly wrote to this address. In 0.9.10 however, since I am writing 4 bytes, this code would be written starting at address 021E61C4 which, of course, screws everything up. A similar argument can be made for the second half of the code where, because I am writing 2 bytes to memory, I start writing at address 021E61CA. With that, I have created a new code for Test2 which works for both DeSmuME 0.9.7, 0.9.10 and no$gba below


 * 021E61C4 547E0072 021E61C8 32747365 221E61CC 00000000


 * Interestingly, from what I have tested, most (if not all) of the US versions of each code still work in every emulator I have tried (DeSmuME 0.9.7, 0.9.10 and no$gba version 2.7a). Its all because the memory addresses that are accessed by each code sit on an address that is a multiple of 2 or 4. Good thing since that cuts my work in half. I guess I will need to update the Japanese codes at some point. Still no idea if the US codes or if the code above works on the real hardware though --Cuber456 (talk)
 * Again, from the programmers:
 * "the concept of alignment universally applies solely to values larger than bytes. so, it seems, theres no evidence that AR supports unaligned word-length patches (in fact, it was always highly unlikely)."
 * So yes, the original codes wouldn't have worked on actual hardware. If you'd like to make new codes (or I could give it a shot) then that would be very much appreciated. -Einstein95 (talk) 15:26, 20 December 2013 (EST)
 * From what I can tell (unless I am dense which is possible and I apologize if that is the case), I don't think the original Japanese codes would have worked on real hardware. However, the US codes, I think, would work on real hardware because those codes are 2 and 4 byte aligned. I think the same is true for the new AR code I created above for the Japanese version. The only way these codes wouldn't work is if real ARs always (and only) use 4 byte alignment. Recreating the codes wouldn't be hard, just tedious since there are so many of them on the main page and notes page. Anybody can create the codes but I'd feel bad if I just passed off the work to someone else. So I'm totally willing to recreate codes, but tell me this, are AR codes always 4 byte aligned??? Would the Japanese code I created above not work because it is both 2 and 4 byte aligned??? Like the saying goes, measure twice, cut once. I want to make sure that I recreate the codes right on the first try. --Cuber456 (talk)
 * Sorry, I haven't been on here in weeks. But thank you all for the tips. I forgot to mention that I tried pretty much all of the ending codes, only to read here that none of the codes do not work on 0.9.8, I should download 0.9.7. Thank you.