Posted by NetSplit

---

If it's of any help, here's what I have in my notes about checkpoints:

$153B0 - Checkpoint scroll byte numbers
$153D3 - Table of values defining the slot number of the last enemy on the checkpoint page (start, mid, end)
$1540A - A table containing the first screen ID in the scroll byte selected for the checkpoint
$15422 - A table containing the last screen ID in the scroll byte selected for the checkpoint?
$1C2E4 - Screen numbers which activate the checkpoint
$1C2FB - Where Mega Man respawns after dying
$1C52A - Checkpoint y coordinates

These tables should be a multiple of #$0C bytes long, with 1 entry for 12 stage slots in order, and up to one set of 12 entries for each of the 3 kinds of checkpoints, in order. Not every table has a set for every checkpoint; for example, the table at $15422 doesn't have data for the startpoint. Setting the data in all of these tables properly should be enough to make edited checkpoints act correctly.

Edit: I updated the table for enemy slot numbers to include the startpoint data; the address previously started at the midpoint data. For the enemy data, the byte is the slot of the last enemy of the page; if there have been no enemies in the stage so far, that should be FF.


I also seemed to think that changing $150F7 to 4CF690 was a good idea when I wrote this many years ago, but my notes aren't clear on why, and glancing at bisqwit's disassembly doesn't make it immediately obvious. I wrote: "the game calculates the data itself that it needs if you're hacking the ROM. If you don't make this change, it ignores its calculation and uses a table elsewhere in the ROM. Has to do with checkpoints". In the disassembly, I see that all my change does is skip past this chunk of code:

$90E7> A5 AB: LDA LastRestartPointType
$90E9> F0 0B: BEQ + ; $90F6
$90EB> 18: CLC
$90EC> 65 31: ADC CurrentStage
$90EE> AA: TAX
$90EF> BD 0694: LDA Label9412-12,X
$90F2> 85 28: STA CurrentEndScreen
$90F4> E0 08: CPX #$08

I'm betting LastRestartPointType contains #$00, #$0C, or #$18 (Edit: This seems to be the case), representing the 3 checkpoints (start, mid, end). If not 0, it uses it plus the stage number as an index into the table at $15422 (treating the start of the table as $15406 so the midpoint data starts at the actual start of the table). This is used as CurrentEndScreen, so the meaning does match up with the table from my notes. Looking a bit before this code, there is some stuff that does indeed calculate a value for CurrentEndScreen. This strikes me as silly; either they should add an additional 12 bytes to the table and read from that every time, or they should do the calculation every time and do away with the table.

You could try making that change and seeing if it means you don't have to do anything with $15422 anymore. I see two potential problems with this. First, the table may exist because their calculation code does not work properly for mid and end points. Second, there is a compare at the end of this code; I don't know if it's used for anything, but it may have some significance that makes bypassing it a problem (Edit: Following the code briefly in a debugger, it doesn't look relevant anywhere). I'm not planning to dig through the code any further now to see if those actually are problems; I leave it as an exercise to the reader!

Edit: As an additional note, if you're specifying the index in the scroll data for this corridor and the screen at the start of the corridor, you shouldn't need to include the screen the corridor ends on because it's easy to calculate. This table doesn't seem like it has much of a real point.

My curiosity won out and I wound up putting this to the test. At every checkpoint I tried (Cut Man mid and end, Fire Man mid and end, Wily 1 mid and end, Wily 2 mid), the table always yielded the same value calculated by the routine. I'm pretty sure that you can just do that 3 byte assembly change and ignore that one table, but feel free to check up on this yourself.


Good luck to you.

---

Posted by NetSplit

---

Also, in that video, be careful around 1:47, since the way you set up the stage makes it so someone could conceivably scroll onto that screen too low, which would cause them to get stuck in the floor, zip to the right, and die in the pit, which is a cheap death and probably not what you intended.

---

Posted by Zieldak

---

But disabling the flashing effect would be a much better way to do this.

---

$15861: 4C 76 BF
$17F86: B0 01 60 A5 14 29 10 D0 F9 4C 6E 98

$9851:4C 76 BF  JMP $BF76


$BF76:B0 01     BCS +       ; This is the code we clobbered at $9851
-
$BF78:60        RTS
+
$BF79:A5 14     LDA JoyPad0
$BF7B:29 10     AND #$10    ; Check if up is held.
$BF7D:D0 F9     BNE -       ; If it is, we don't want to release the ladder, so exit.
$BF7F:4C 6E 98  JMP Ladder_Release

Posted by NetSplit

---

You can disable the flashing effect by changing $1532D to EA EA.

---