Toggle sidebar

Homebrew game Project [Release] The Minish Cap – Alek's Ultimate NX Edition | Dual Screen, FLIP GRIP SUPPORT, Touch, RetroAchievements

  • Thread starter Thread starter alexgg1014
  • Start date Start date
  • Views Views 11,928
  • Replies Replies 126
  • Likes Likes 16
Thank you. I just have a small suggestion: when you press select for the menu, the options appear first. Could we make it remember the last open tab?
 
  • Like
Reactions: digdat0
flaviometal said:
New found bug: after beat Palace of Winds boss, game crashes. It happened to me twice.

Thanks!
Click to expand...
Alek's Ultimate NX Edition v1.4.14 — Gyorg Pair crash, a dojo crash, and a pass over Vaati

(Edited: this replaces the 1.4.12 and 1.4.13 notes that were here — 1.4.14 includes both of them.)

Available both ways:

Fixed: the Gyorg Pair fight crashed in its final phase (thanks @flaviometal, and mnavarretem86 on GitHub for the crash log and save)
It closed the game when you hit the exposed eyes, which is also the killing blow, so the fight couldn't be finished. Each Gyorg male removes itself from the boss's shared data the moment it finishes dying, so by the last phase there is none left — and the eye code was still going looking for one and reading through the empty slot. It now treats "no male left" as what it means: defeated. The crash log pinned this to the exact instruction, so this one is confirmed fixed.

Fixed: entering a dojo could close the game
Nobody reported this one — it turned up while I was regression-testing the release with the save from the Gyorg report, and it had been there for several versions. When a room's light is already off, the light manager runs outside the normal entity update, and in that path it was deleting whatever entity happened to run last instead of itself, or nothing at all, which is what crashed.

Also: a hardening pass over Vaati and 24 more files
Since people are getting to the end of the game, I went through the final boss the same way. Two of its parts were walking their shared data with byte offsets that only add up on the GBA, which on this port lands in the wrong place and can overwrite a pointer to another part; the rest are checks for parts that failed to appear or that die before the code looking for them.

Being honest about this part: no Vaati crash has been reported and I couldn't make one happen — the arenas behave the same before and after, and the screenshot set I compare every build against is pixel-identical. These are guards put in before the reports arrive. A few bosses (Mazaal, Moldorm, Gleerok) are still pending because their fixes need machinery this port doesn't have yet.

Hitting a crash? Attach the sdmc:/switch/tmc/crashlogs/ folder (or the newest crash_*.log) and tmc.log — those carry the fault address and the exact build, which is what pins down the line.

SHA-256 a40d252e52e07bd5187bb21358f9220fb683f307f59595a71d6a2cb535a3fca1 — tmc_aleks_ultimate_nx.nro (14,223,027 bytes)
Post automatically merged:

Alek's Ultimate NX Edition v1.4.16 — four-cannon puzzle fixed, and items on X / Y / ZL / ZR from the pause screen

Available both ways:

Fixed: the four-cannon puzzle in Dark Hyrule Castle couldn't be completed (thanks mnavarretem86 on GitHub for the video and the save)
All four Links reflect their cannonball, but only the two statues on the left were ever destroyed. The reflected ball looks up the statues through a shortcut that lands in the right place only on the GBA; here it started 16 bytes early, so of the four slots it read, two were unrelated data and only two were real statues — always the same two. It reads the real list now.

Verified by measuring the offsets in the build (the shortcut points at +0x28, the statue list is at +0x38, which is exactly the two-of-four you see in the video), but I couldn't drive the whole puzzle myself, so if that room still misbehaves please say so.

New: put items on X / Y / ZL / ZR straight from the pause screen (suggested by lbsbezerra on GitHub)
Open the game's own inventory, highlight an item and press X, Y, ZL or ZR — it goes on that button, the same way A and B work in vanilla. The second screen's ITEMS tab still does it too, and both views agree.

The other half of that request, drawing the X/Y buttons in the in-game HUD, isn't in yet: that HUD is the GBA's own tile-built one and already moves around in widescreen, so it needs actual artwork rather than a hook. If anyone has the ABXY mockups from the Picori Discord, I'd love a link.

If you're on 1.4.14 or older this also includes the dojo crash, the Gyorg Pair final phase and the Vaati hardening pass.

Hitting a crash? Attach the sdmc:/switch/tmc/crashlogs/ folder (or the newest crash_*.log) and tmc.log — those carry the fault address and the exact build, which is what pins down the line.

SHA-256 47bd60d9b4de0b439dc4aad92614d00615b4225abcc98b98172eaa4dfbc813df — tmc_aleks_ultimate_nx.nro (14,223,027 bytes)
 
Last edited by alexgg1014,
  • Like
Reactions: JJTapia19 and digdat0
Alek's Ultimate NX Edition v1.4.18 — the final battle is beatable again, plus the X/Y/ZL/ZR hotfix

Available both ways:

Fixed: the Dark Hyrule Castle final battle skipped Vaati Reborn, and Vaati Transfigured started over every time you beat it (thanks mnavarretem86 on GitHub for the report, the video and the save)
Both came from the same line. The port carries a safety check that finishes off Vaati Reborn when his phase counter is already past the last phase — it's there for a boss restored from an old quick save. It ran from the boss's very first frame, and the boss is spawned with the room and sits idle until the intro cutscene sets him up, so until that moment the counter holds leftovers from whatever used that entity slot before. On the reporter's save it read 56, which is well past the last phase, so the check fired on frame one and ran the defeat sequence immediately.

That one misfire causes both things you see. The defeat sequence raises the flag the "Vaati defeated" cutscene is waiting for, so you get sent straight into Vaati Transfigured and the Reborn intro never finishes — which is why the game never records that you did phase 1. Come back after beating Transfigured and the same misfire happens again before the boss can remove himself, so he dies again, raises the flag again, and you land back in Transfigured. That's the loop.

The check now only runs once the boss is actually in the fight. If your file is stuck in that loop it recovers by itself on this build. If the Vaati-and-Zelda scene never registered on your file, walk back down to the Zelda statue platform — that scene replays when its flag is clear and the sequence picks up from there.

Also fixed (1.4.17): assigning an item to X / Y / ZL / ZR from the pause screen changed what was on A and B
The shortcut added in 1.4.16 did its job and then equipped the same item a second time: a button holding an item forces the GBA's B internally, which is how the port fires it at all, and inside the inventory that press is vanilla's "equip to B". With the cursor on the sword the two boxes even swapped places, so it looked like X wrote A while Y, ZL and ZR wrote B. One bug, four symptoms. The press now only assigns, and holding the button while you close the menu no longer uses the item on the way out.

Not in yet: the X/Y buttons drawn in the in-game HUD
Still the other half of lbsbezerra's request, and it's coming in 1.5. That HUD is the GBA's own: the button is a real sprite whose tiles live in a fixed, packed block of video memory, so adding two more means reserving space there properly rather than grabbing a slot that looks free — an earlier attempt in this port did exactly that and leaked into the pause menu's sprites. Doing it the right way is the work. Artwork for the buttons is already done (thanks lbsbezerra); ZL and ZR still need theirs.

If you're on 1.4.15 or older this also includes putting items on X / Y / ZL / ZR from the pause screen, the four-cannon puzzle in Dark Hyrule Castle, the dojo crash and the Gyorg Pair final phase.

Hitting a crash? Attach the sdmc:/switch/tmc/crashlogs/ folder (or the newest crash_*.log) and tmc.log — those carry the fault address and the exact build, which is what pins down the line.

SHA-256 d76f16b2386615788a1becfb4ab4e9ef16f49ea80665958aaa9ac6c22c6889a7 — tmc_aleks_ultimate_nx.nro (14,223,027 bytes)
 

Site & Scene News

Go to forum More news

Popular threads in this forum