JoypadLowSensitivity

[10:28 AM] entrpntr: hmmm, I am growing suspicious of the naming screen
[10:28 AM] entrpntr: I started double checking to make sure I was actually getting through the textboxes properly
[10:29 AM] entrpntr: and most of the time I get there on the correct frame and RNG
[10:29 AM] entrpntr: but sometimes I come away with DVs not on the list
[10:29 AM] entrpntr: hra/hrs off by 3ish
[10:29 AM] entrpntr: B212 instead of B20F
[10:30 AM] entrpntr: FE6E instead of FB6B
[10:30 AM] entrpntr: but occasionally I do get something on the list
[10:31 AM] entrpntr: this is using the same exact save between bot and BGB
[10:33 AM] entrpntr: nothing I do to alter the inputs seems to matter, so I'm stumped for now
[10:37 AM] entrpntr: 86D1 instead of 81CC
[10:37 AM] entrpntr: etc.
[10:41 AM] entrpntr: it seems like if the A and START are spaced out far enough, then bot gets what BGB gets
[10:58 AM] entrpntr: both bot and BGB appear to be fixed in what they come up with no matter how I alter things, so this seems to me like emu difference
[11:22 AM] entrpntr: ok now I somehow forced FEFD
[11:22 AM] entrpntr: on BGB
[11:26 AM] entrpntr: while testing a savestate for FFFE audio cue
[11:26 AM] entrpntr: got FFFE like 25 times, then randomly an FEFD
[11:26 AM] entrpntr: so there is something subtle going on I guess
[11:42 AM] entrpntr: and now forced 00FF somehow
[11:43 AM] entrpntr: so there are definitely cycles that haven't been accounted for
—
[9:12 PM] entrpntr: haven't fooled around much, but you can mess with the RNG in naming screen by typing extra characters then erasing them
[9:12 PM] entrpntr: I was getting 01FF when typing 2, deleting 1 in BGB, bot thinks it should be 00FF
[9:13 PM] entrpntr: idk if there is just subframe dependence, so cycle timing of inputs produces the random failures
[9:14 PM] entrpntr: or if there is an explanation which can make things preventable
[9:50 PM] Exarion: does it matter which character(s)?
[9:51 PM] Exarion: could explain gunner's issues (he names B, and I think everyone else does A)
[9:51 PM] entrpntr: I haven't tried anything with the dpad thus far
[9:52 PM] entrpntr: regardless, it seems like anything that goes to the naming screen has a chance of failing from what I've tested on Squirtle
—
[11:37 AM] entrpntr: I haven't really tested much more on the naming screen, but I was going to look at messing with inputs on frames where IGT overflows to see if that is somehow it
[11:37 AM] entrpntr: or if it is printing characters or something else on those frames that causes the offset
[11:38 AM] entrpntr: from minimal testing, buffering the dpad on the naming screen didn't affect anything at all
[10:44 PM] entrpntr: ok so I have definitely confirmed the problem on the naming screen
[10:44 PM] entrpntr: frames where you press and release A both seem to matter
[11:28 PM] entrpntr: nvm just the frames you press A (or B) matter
[11:33 PM] entrpntr: https://pastebin.com/VNZSbTc2
[11:35 PM] entrpntr: yeah that looks like JoypadLowSensitivity
[11:36 PM] entrpntr: failures are in clusters of 5 frames
[11:38 PM] entrpntr: right, because you can't reselect A or B within 5 frames?
[11:39 PM] entrpntr: so different code flow
[11:46 PM] entrpntr: and it indeed looks like buffered dpad is fine, so B and I should work for Nido
[2:46 AM] entrpntr: I'm pretty sure IGT was just a red herring for the naming screen, all fuckery is from JoypadLowSensitivity
[2:50 AM] Exarion: so timing or duration of the A press?
[2:50 AM] entrpntr: timing
[2:53 AM] Exarion: what does that have to do with sensitivity
[2:54 AM] entrpntr: extra code will run before Random calls depending on when you press A
[2:55 AM] entrpntr: there is a 30 frame counter and a 5 frame counter
[2:56 AM] entrpntr: which produce some interesting patterns
[2:58 AM] entrpntr: magnitude of fuckery can vary, probably something that depends on rdiv alignment on certain frames
[2:58 AM] entrpntr: but an A press exactly 5 frames before Start press seems to always be pretty hard