If you want real bugs, watch this instead:

https://www.twitch.tv/andy/v/98104910

jump to 21hours 6minutes and 30 seconds and wait for it!

This also happens 2 minutes later at 21:11:00.
I tried to reproduce hard but didn't manage. He uses a hdma item while bouncing back it seems.
Also Skuj's room bug I haven't been able to reproduce. Maybe this is due to that he hovered with the boots across the gap. There's really some weird gameplay this guy is doing.

I must wait for Euclid's opinion, I cannot fix what I cannot reproduce.. In the meanwhile I check the bat.

The HDMA crashing bug probably happens as a result of using multiple drugs in the same screen, kinda like Erock got a few weeks ago:

Erockbrox on October 18th wrote:Here I found a game crashing bug. I was trying to see if there were any errors with using multiple items that use custom HDMA such as the drugs. After using the speed I used my dash boots and somehow when going off screen I caused a crash.

I'm thinking it might be that I was turning on a screen boundary with the boots or something like that.

The music still played and I could hear the R and L buttons switching items, but everything else was broke.

This happened in pretty much the same way as Andy. Taking several drugs and bouncing against an object which would then trigger the next screen loading. So the boots bouncing might have something to do with those unusual crashes!

Will check once more, but have little hope.
Edit: No, I can't  


I now know why the bat animation that opens the batcave doesn't work anymore, the sprit's x,y coordinates were changed.

E.g.,
current version:
$9A/F71B A9 07       LDA #$07
$9A/F71D 99 30 0D    STA $0D30,y[$80:0D3F]  

old working version (e.g., StargateZ3v6900.smc):
$9A/F71B A9 09       LDA #$09                
$9A/F71D 99 30 0D    STA $0D30,y[$80:0D3F]

But there are also some other bytes changed. Restoring them will let the bat reappear, but not the complete animation (
$9A/F71B A9 09       LDA #$07 -> 09
$9A/F716 A9 4C       LDA #$CC -> 4C;
$9A/F725 A9 06       LDA #$06  -> 0a)

A few Routines above, there is a JSL to some of Euclid's code:
$8D/B829 22 00 D0 45 JSL $45D000[$45:D000]

I do not know what it is for and why the coordinates of the bat were changed. I assume this may have something to do with Ganon?

In any case, here's a savestate... Since I do not know what Euclid has done here I do not dare try to fix anything. Euclid, do you have time to check what's wrong? You can check with the savestate the current and an older version, as said, v6900 works).

whatever it is its not the boss patches, the 45/d000 routine is for hp and i havent touched bank 9a/1a

I'll check them tonight, i did tweak agahnim's x/y for a different reason

Simpsons rom i have for beta test 7162 works with the bat (that has all boss patches) - something applied afterwards...

the only patch which should've been applied are the hex edits for the pond + header, which doesn't affect bank 9a/1a.

What was applied after that? i saw a crapton of bytes changed around that area including some un-fastrom patch (i saw a jsl went from 9b -> 1b)

In this case I assume that a HM sprite edit by SePH might have caused this? HM sometimes changes some bytes.
After the Simpson patch I only did those minor bug fixes, and also didn't touch fastrom as well. Or did you apply a patch neither made by Euclid nor me, SePH?

If you want to help shed some light into this you can check your backups when
0D/771C was changed from 09 -> 07 (09 working, 07 not working)
I think you have a good documentation and can tell what was done breaking this animation.

I dug into the HDMA bug a bit.

It's caused by movement speed.

Here's some easy to reproduce steps
1. go to that same spot behind sony substation
2. dash into the substation, stay there
3. take speed (link should "twitch a bit" - this indicates you've done it right)
4. now dash and turn direction to north - you should bounce off the bottom the screen.
5. move again - transition occurs.
6. crash.

Unfortunately changing the moving speed is something introduced in conker and is literally breaking lttp's engine. Nothing I can do about it.

NB reproducible with beer, not with weed (weed slows you down so the pixel offset doesn't cause link to have scrolling error)

I'll leave this patch here with this gigantic warning.

I undid a region of bytes around the area Conn pointed out from the Simpson rom I have - loaded up the save state from Conn and the bat is back.

I have not done a full play test over this rom and do not endorse apply this patch at all. Use this as a last resort and there needs to be a full play test of the entire game prior to saying it did not undo any required fixes

Had some thoughts with regards to that dash bug - if you can restrict the boundary which conker bounces back (i.e. not too close to the edge of the screen) then the bug shouldn't occur

patch works like a charm

But you are right in any case, as long as we do not know what action broke the animation in first place, it is unsafe to apply this patch without a complete playtest. I hope SePH can shed light in this. If HM is responsible, I think applying the patch, make a safe in HM, may reverse your byte changes again.

I'll also try to reproduce your receip to make this bug occur. But this is a bug in any case 99.9999% of people won't run into. If it isn't fixable, I think this isn't a major problem.

Regarding the crash, I could reproduce it and Euclid is absolutely right. If you bounce back to the border of a screen, this will happen, and I also found no asm solution for this. Possible fixes:

- design: shift all obstacles close to a screenchanging border (sony substation, that Hyrule castle rail where Andy caused this bug), so that you cannot bounce to a screen border
- asm: make the bounce back not cause any shift in the x/y coordinates of Conker (so that you stay at the coordinates where you hit the obstacle).

Design change will be no option I guess since this will require more work than you want to spend I guess. Also, I do not know whether also dungeons are affected.

ASM change will have the disadvantage that some puzzles will need to be redesigned. But since you get the feather quite early I think you do not have such puzzles...? So this is the most obvious solution.

I don't think it affects dungeons at all, so maybe the fix is to round off a few more numbers when bouncing to the edge of the screen in the overworlds.

Though I hate to be toying with that collusion code at this stage.

Try this:

disable left rebounce:
03/91C3: 18 -> 00

disable right rebounce:
03/91C4: E8 -> 00

disable up rebounce:
03/91BD: 18 -> 00

disable down rebounce:
03/91BE: E8 -> 00

I think this is the easiest solution.

the negative value E8 must be counted up while the positive value 18 counted down to make the rebounce distance smaller, just ensure you count e8 with the same value up as 18 down, like
E8+8=f0
18-8=10
This will decrease the distance by 33%, if you do not want to disable it completely.
Accordingly, f8 and 08 will reduce it by 66%

I don't think any of the puzzles requires you to bounce back against walls, this is more something speedrunners use most of the time, but if it can fix the glitch, might as well implement it.

So should I apply Euclid patch as well or include it separetely from 1.1 along with some note that the patch needs further testing?

Want to be sure.

Also superskuj has made a sprite editor that allows us to change the prize packs and many other sprite settings, so I might move 1.1 further in order to fully optimize by game a bit more!

He will release it on Zeldix once he feels like it is complete!

Ok, in this case I suggest you disable the bounce back or decrease it by 66% (08 and f8).

As for Euclid bat patch, I do not know. I hope you can find out the change caused this. Neither me nor Euclid changed anything here in first place.

Still unclear remains that feather bug, he encountered in skuj room. I can only imagine that his hoovering above the gap may have caused this. Any ideas?

More time is also great giving me the possibility to adjust the slide times. Qwerty pm:ed me, he's done with the tracks at weekend.

Can I asm the bounce back for overworlds only? i think Erock's puzzle needs the bounce with his bouncer in front of gap room.

I can probably fit a full playthrough over the weekend with msu, is there any other fixes? Aside from:
- conn's fix asm
- the bat fix
- the outdoor bounce
- last bit of msu changes? (where is it?)

I'm interested in anything which got applied after the simpsons rom you gave me.

Good thinking with the rebounce! I remember you need to wear the ring while rebouncing at one puzzle in Erock's room and this is thus not solveable with the feather.

- last bit of msu changes? (where is it?)

That will be just lengthen the ending slides. Erock has made spoken text of the text on those slides, qwerty will made the ending themes with voice overs. He already tried but the 14 seconds each slide has is just too short for some of the text, so I will adjust their length individually. This is just a byte change for each slide to match.

If you test your bat fix you probably should test a safe in HM first after applying the patch before starting the walkthrough. If it was caused by a sprite change done with HM, HM surely will reverse your fixes again.

Any help needed by me other than the end track length?

Yeah bounce back overworlds only might be a good idea. Forgot about those Erock puzzles.

Bounce fix.

Fun fact: if you manage to mysteriously bounce yourself on the edge of the overworld screens from dashing (happens a lot during speed) - you may have just reproduced the crash bug.

So I'm going to
1. take the public release
2. apply conker_final_v2 from https://www.zeldix.net/t1245-bugs-and-fixes
3. apply the bat fix.
4. apply this no bounce patch

and play through the game as quickly as I could to get ending 4 with all items on msu and report back in about 18 hours time (sleep, then housework, then play)

What I wouldn't have done are
1. fix the crc (meh)
2. lengthen the end screens timer
3. have the new msus (I peeked on his private cloud server and it has ending 1/2 uploaded along with the lamolas pause, but nothing else)

PS: tinypic and photobucket is bit weird on the site for me so it's a bit slow.

These don't need to be fixed - just general observations

Minor stuff from this testing (may already be existing bugs):
- wall moving sound disappeared from Ganon's Lair after lighting torches false alarm - it's due to sfx stacking.
- slot machines don't seem to work the first time you get into the room, coming back afterwards does though.
- Golden Lighter - if you continuously use it, a different graphics come out the second time you use it. (only as soon as you got it in the mansion)

Finished!

All good! Don't try to edit in HM anymore after applying these patches - I have a feeling it might be changing the entrance for Chris Houligan that did it.

So for 1.1 do the following:
1. take the public release
2. apply conker_final_v2 from https://www.zeldix.net/t1245-bugs-and-fixes
3. apply the bat fix.
4. apply the no bounce patch
5. lengthen the end screens timer (if required)
6. Edit 0xE6E55 from 0x34 to 0x35 (this changes the monologue from 1.0 to 1.1 on your sign)
6.1 Vaati fix (it's safe enough to use)
6.2 Apply your preferred underverse boss rush respawn patch
7. fix the crc
8. fix the readme

Then call it 1.1

I'll also probably tweak some enemies and prize packs, but dont think that will affect other things.

Also...

https://clips.twitch.tv/andy/InexpensiveHerringKAPOW

LOL, that damn Andy did it again!

That's probably a bug from original lttp - not in charge of fixing original lttp bugs...

It doesn't crash the game like that bouncing one.

So tinkering Chris Houlians room was responsible for the bat vanishing? I hope it didn't affect other stuff.

I could reproduce Vaati, this happens when you lure vaati to the very right or left border of the screen (doesn't happen up/down), like in Andy's twitch. Watching $0d30, the children sprite's x-ccordinates, the  values are staying out of screen, in the right room, therefore they vanished.

Possible fixes: make the room smaller (move the right wall more to left, and the left wall more to the right). Maybe it is also possible to restrict Vaati to not be able to get to close to the left/right border? I noticed that he cannot go completely up.

An asm solution in any case isn't possible, unless Euclid depimps him and restores the original smaller radius (which in turn makes him easier but probably fixes this bug).

i can probably do the same to him as bowser - its the same routine. Though i'd be more happy to call it an easter egg.

I have noted kholdstare doesnt die properly if you hit it with the goldstar or something before the split animation finishes.

Someone pretends its possible to do that in alttp as well so wouldnt bother with it and leave it.

Dont think people will trigger the kholdstare one...