Thanks to all for the kind words, it helps me get more motivated!
Yeah, Photon has been the main (?) coder on 'The Player 6.x' the last couple of years... With all respect to The Player, I still think that it's too inaccurate in some situations, but I might be crazy about small details. Not sure if it can do 128K samples properly.
PT2.3E's .MOD format is identical to that of PT2.3D and older PT versions. You can find the PT1.x MOD format documentation all over the net.
I haven't changed anything in the format structure, though I have fixed one bug in the module replayer so that sample loop points above the 64K barrier play correctly in the voice handler. This means that if you were to use such loop points on a >64K sample in a PT2.3E/PT3.62 MOD, it will not play correctly in PT1.x/2.x/3.15! PT3.62 had this fixed, so this is not the first PT version to have that replayer difference.
It was sad having to touch the replayer code, but I felt that it was an essential compromise because I've fixed probably all 128K sample bugs in the whole program (non-replayer stuff). Having 128K samples is already possible in the PT MOD format without hacking it: sample loop points and lengths are stored as words (for directly feeding them to the Paula regs I guess), not bytes, so 0xFFFF would result in 0x1FFFE bytes.
Last edited by 8bitbubsy; 27 January 2016 at 12:09.