JoeGtake2 wrote:
Out of curiosity, did you by chance navigate away from the window of the game and return at any time?
No. Windowed mode in your build of GGVM crashes immediately for me, so I just played it in fullscreen. Trying again today, though, I managed to reproduce it reliably in FCEUX, so you can rule out GGVM.
I wrote $FF to $73D to give myself "lots" of notes. (Curiously it seems to count as 999, but maybe there's some decimal overflow conversion going on, I dunno.) After doing this, it seems I can randomly spam notes for a while (a few hundred rapidly played notes) and eventually "weird" stuff happens. I think the first time it happened to me I only had about 50 notes but I got "lucky".
Take a look at the attached FM2 file for FCEUX (used with game.nes from the .jar). The player palette changes and loses the staff, I think at frame 1496 of the movie. From there you can walk to the next room and you'll see that counters for notes, coins, and EXP are all corrupted. If I keep spamming for a while, I've also managed to make the colour of the world change and dying after that took me to the "10101010" zone again.
You can use the debugger around frame 1495/6 in that movie to inspect what's going on.
Another bug I mentioned earlier that I whacked a monster a bunch of times on the right side of the screen. I think the right side is a red herring here, later I was fighting a different monster and they seemed to get stuck in the "taking damage" state (different palette, couldn't be hit by it again). Eventually it came out of this state though. Maybe the timer on that state can sometimes get decremented twice to -1 by accident or something?
JoeGtake2 wrote:
We do have the occasional sticking, especially against jumpable objects. it's not quite game breaking, because you can jump to get unstuck, but it is definitely something we'll be fixing.
I think you misunderstood what I meant about sticky walls. This isn't occasional, it's systemic. If you walk left into any wall, you can't move up or down until you release left. (Same is true for any direction of motion.) The initial collision prevents valid motion on the other axis.
If you need a better illustration of what I mean, try playing Link to the Past (SNES) and walk diagonally into any wall, and compare that to doing the same in your game.
JoeGtake2 wrote:
shop keep...that is as intended, so you can *exit* an npc.
It's actually a big problem for critical events that are tied to the end of a dialog.
For example, the wizard at the end has 6 pages of dialog. If I press B even slightly before the end of any of these 6 pages, the whole dialog is cancelled, and the "ending" is not triggered.
There's no feedback to indicate there is more dialog. Unless you know that dialog doesn't continue unless you wait, it's very easy to end a dialog early (by pressing slightly before the end of a page) and think you've seen all of it. I could easily end up thinking "I talked to that guy and nothing happened", and being stuck because there's a critical event tied to finishing that dialog.
I'd highly recommend putting dialog cancel on a different button than dialog advance. The way it is now, pressing B slightly before the end of a page displays, or slightly after, has the opposite effect!!! Kind of a strong difference to hang on such a subtle timing difference!
I had a lot of problems with the cancelling because I was confused by it. I ended up trying to press B to make dialog go faster, but it just kept cancelling things, and I didn't understand what was going on. I think impatient people will try to press the "advance" button rapidly like I did, so I'd really recommend it speed up the text instead (and put cancel on a separate input), or if speeding up the text is impossible it would be fine even to just do nothing until the page is finished and you are allowed to advance.
JoeGtake2 wrote:
Did you eventually see the win-screen?
Yes, if you mean the "coming soon" screen, looking out from the mouth of a cave.