Awesome! Finally, something that actually feels like progress! I'll check your findings, and see if anything else pops out at me. Seems like every time I take a nap, you guys figure more out without me, lol. Maybe I'll just hibernate for a week (j/k)
Anything I find will be edited into this post, so for those following along, check back here in, say, a year or three (again, kidding... don't know how long it will be though)
[UPDATES SO FAR]
Confirmed SeaofTea's observation about g6dsload.1 (fe ff = Sakura setting, fe fe = Touchpod)
Also confirmed the XOR trick on the headers - also appears to work on regular touchpod files (not fully tested yet)
To add to the list of conversions,
Eng to GB(Asian) = XOR 0x31
These conversions work both ways, of course, i.e. XORing jp files by 15 twice in a row will restore them to original (Any decent programmer should already know that, but for those of you who are experimenting, thought that might help)
File comparisons (work in progress):
homebrew.eng is identical between the Asian Sakura and Asian touchpod, yet different from the english touchpod - further analysis is needed
While comparing menu.jp between 1.34 and 1.12 (unhacked) Sakura, I've noticed a number of places where the order of 8 bytes has been reversed - swapping the first 4 bytes with the second 4
i.e. 11 22 33 44 55 66 77 88 becomes 55 66 77 88 11 22 33 44
@SeaofTea - I checked out those Robert codes and noticed more to the pattern. In the original files, each line starts with:
08 40 2d e9 xx xx a0 e3
The modified files have a similar pattern:
a4 xx a0 e3
(I used 'xx' in place of bits that aren't pertinent, although they are all the same in the modified files, they vary in the originals)
Instead of 8 parts, there now appear to be 11 in startnds.jp that match this pattern
So it looks like the xx xx part might be important, some sort of address maybe, while the 08 40 2d e9 might be the region code... though why our region code would be one byte compared to their 4 bytes, I can't begin to guess. I'm still studying the differences. Going to try something soon, updates to follow
Also, the ROBERT part seems to be unimportant, maybe it's detected as text somewhere in the code - And I'm guessing that iamnobody's real name might be Robert
The four bytes preceding 'ROBERT' seem to go with this, as they aren't all the same in the original, though they are in the modified version. Perhaps they are formatting codes?