1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
|
# Movement feedback, round 1
The first round of playtest feedback on movement, and the changes we made in response. The numbers
here are as of this round. `src/config.h` is the source of truth, and `game_feel.md` holds the full
feel guide and tuning history.
---
## What players reported
1. **The second jump's timing was too hard.** Players pressed A a hair early, while still falling,
and the press was lost: no jump happened on landing, so the second-jump window slipped by.
2. **Players expected to wall hang on a single jump.** Only the second jump, retreat jump and wall
jumps could grab walls, so running at a wall and jumping once did nothing.
Two smaller requests came with them:
3. Stopping should be very slightly quicker when the stick is released. The skid stays as it is.
4. Starting a skid (`retreating_movement`) should need a slightly more deliberate pull-back.
---
## What changed
### 1. A 0.1 s jump buffer (the second-jump timing)
An A press is now remembered for `PLAYER_JUMP_BUFFER` (0.1 s). The first moment a jump is possible
within that time, it happens.
- **On the ground** it's instant, just as before.
- **Pressed while falling**, it fires on the landing tick. A press up to 6 ticks (0.10 s) before
landing still gets the second jump; 7 or more ticks early still does nothing.
- **Pressed just before grabbing a wall**, it fires an early wall jump the moment you grab.
- Holding A still decides hop vs. full jump. A buffered press that was already released gives a hop.
Implementation: `Player.jump_buffer`, set in `player_update` on `jump_pressed` and counted down each
tick. Jumps start through `take_jump()`, which spends the buffer, in the ground, skid and wall-hang
states.
### 2. Every jump can wall hang
`heightened_movement` now turns on when **any** jump starts, including a plain first jump, and lasts
the whole jump. Landing still opens the usual 0.2 s second-jump window. A single jump into a violet
wall now grabs it.
**A knock-on fix: the clear-the-top rule.** Once every jump could grab, hopping onto a ledge from just
below its lip would grab the wall face instead of landing on top. That's especially bad on the
hangable cliff. So now you don't grab a wall whose top your jump will carry you over anyway
(`would_clear_top()`: apex = y + vy² / 2g, compared with the wall's top).
- A normal jump at the cliff from 1.2 m under its lip → lands on top.
- The same jump 10 m below the lip → grabs the face.
This also fixed an older snag, where a second jump from just under the cliff lip grabbed the face.
To go back to strict "always grab", remove the one condition in `wants_wall_hang()`.
### 3. A slightly quicker stop
`PLAYER_GROUND_FRICTION`: 11.3 → **12.0** m/s² (about 6%).
| From full speed (6 m/s) | Before | After |
|---|---|---|
| Stop time | 32 ticks (0.53 s) | 31 ticks (0.52 s) |
| Stop distance | 1.54 m | 1.45 m |
The skid (`retreating_movement`) is unchanged.
### 4. A more deliberate skid
`PLAYER_SKID_ANGLE_DEG`: 120° → **130°**. You have to pull the stick more squarely back to skid.
| Pull-back from the run direction | 120° | 125° | 130° | 135° | 140° |
|---|---|---|---|---|---|
| Result | run | run | run | **skid** | **skid** |
---
## What didn't change
Measured and identical to before: the jump heights (2.33 m / 4.08 m second / 4.67 m retreat), the
0.2 s second-jump window, the skid (0.5 s glide-then-plant), the retreat jump, wall hang timing
(0.5 s, early wall jump in the first 0.1 s), mirrored wall jumps, the one-wall lockout, chimney
chaining, and riding moving platforms. The cliff course is still completable.
---
## Things to watch in the next playtest
- **Wall grabs are now everywhere.** Any jump into a violet wall or hangable rock grabs it. Near the
cliff and pillars that could feel sticky, or it could feel great. Watch for players getting
caught on walls they meant to jump past.
- **The clear-the-top threshold** is exact (apex vs. wall top). If players still snag on lips they
*nearly* clear, a small margin (e.g. grab only if the apex is at least 0.2 m below the top) would
help.
- **The buffer length.** 0.1 s fixes early presses in tests. If second jumps still feel strict,
0.12–0.15 s is the next step. Going much longer risks jumps firing when players didn't mean them.
- **The skid angle.** 130° is stricter. If skids stop happening when players want them, back off
toward 125°.
---
## Follow-up after playing it
### Reverted: the first jump no longer wall hangs
Change 2 went too far: **wall hangs off a single jump made them too strong.** A plain first jump no
longer turns on `heightened_movement` (it's back to the second, retreat and wall jumps, and wall
hangs). A first jump into the main wall no longer grabs it.
**Kept:** the jump buffer (1), the quicker stop (3), the 130° skid angle (4), and the
**clear-the-top rule**. The rule was added for change 2, but it still earns its place: a second jump
from just under a ledge's lip lands on top instead of snagging on the face.
### Cliff course regenerated
Feedback: the platforms were too small (skidding off them), too crowded against the cliff, too
many needed double jumps (sometimes two in a row), and it wasn't clear which gaps needed a double.
| | Before | After |
|---|---|---|
| Platform size | 3–4.5 m | **4–6 m** (all 39 measured in range) |
| Distance from the face | hugging it (0–9 m) | **1.5–13 m** (climb average 6.2 m); only the last few come in, for the hop onto the top |
| Heights | any value | **on a 1.5 m tier grid** from the cliff top: one tier up = a single jump, two = a double |
| Route steps (current seed) | mixed rises, doubles common | **19 singles, 8 doubles, 1 level; 0 back-to-back doubles** |
| Extra platforms | random spots and heights | each **one single hop beside a route platform**, level or one tier up or down |
| Rock pillars | placed first (and blocked the route) | placed after the route, never blocking one of its hops, 4–6 m out |
| Far cliff (the finish) | 60 m | **58 m**, two tiers above the top, so it's on the grid |
| Platforms in total | 37 | 39 (7 moving) |
To fit a mostly-single-jump climb onto the cliff, the route **switches rows** (out a row, or back
in, at the same spot along the face) more often when it's getting too far along for its height.
The course test confirms that the top and the finish are reachable, that the bot lands every
still-platform hop it tried (30 of 30), and that all 7 moving platforms can be ridden.
---
## Files touched
`src/config.h` (new `PLAYER_JUMP_BUFFER`, friction, skid angle), `src/player.c` / `src/player.h`
(buffer, first-jump heightened and its revert, clear-the-top rule), `src/level.c` (the regenerated
cliff course: tiers, bigger platforms, row switches, pillars after the route, far cliff at 58 m),
`README.md` (controls table), and `game_feel.md` (rules, cliff course, tuning history).
|