The roadmap issue W7 calls the drop a "sky barge and wing-sail", and that's what is built: a carrier crosses the island at 150 m on a seeded straight route, and players leap, fall, open a sail and steer down. The GDD's core loop (written in F2) describes it differently: "A packet ship sails a straight line past the island. Each player picks a moment to leave by lifeline mortar and steers a gull-kite to the ground."
Only names and visuals depend on this. The simulation (src/sim/drop.gd, Drop.NONE/ABOARD/FREEFALL/GLIDING) works for either. No client visuals exist yet; they come with the v0.6 art pass, so this needs deciding before then.
Options
A. Sky barge, as built. Simple, and the route crosses land, so every spot on the island can be reached. The risk is that a flying vehicle with gliders leaving it is the genre's most recognisable opening. Under docs/ORIGINALITY.md ("if a player would say that's from that other game, redesign it") it would need a strongly seafaring look: a sail-rigged balloon hull of tarred planks, net floats and rope.
B. Packet ship and lifeline mortar, as the GDD says. A ship sails past offshore and players fire themselves inland on a mortar line, then steer a gull-kite. This is more distinctive and fits the rising-sea identity. It needs the launch to reach the far side of the island: the mortar sets the leap's speed and direction, and the route rule becomes "passes within X m of the coast" instead of "crosses land". About half a day of simulation work, with the same freefall and kite code.
C. Hybrid. Keep the crossing route, but make the carrier a kite-towed sky raft launched from the packet ship. The GDD changes only in names.
Recommendation: B. It's the most original, and the GDD already committed to it. The current mechanics carry over: only the leap's start position and velocity change. If time before Monday is short, ship A's mechanics with C's names and look, and do B after v1.0.
Labelled needs-eric, so the scheduled build leaves this for Eric to decide.
**Found while doing:** W7 (#14)
The roadmap issue W7 calls the drop a "sky barge and wing-sail", and that's what is built: a carrier crosses the island at 150 m on a seeded straight route, and players leap, fall, open a sail and steer down. The GDD's core loop (written in F2) describes it differently: "A packet ship sails a straight line past the island. Each player picks a moment to leave by lifeline mortar and steers a gull-kite to the ground."
Only names and visuals depend on this. The simulation (`src/sim/drop.gd`, `Drop.NONE/ABOARD/FREEFALL/GLIDING`) works for either. No client visuals exist yet; they come with the v0.6 art pass, so this needs deciding before then.
**Options**
- **A. Sky barge, as built.** Simple, and the route crosses land, so every spot on the island can be reached. The risk is that a flying vehicle with gliders leaving it is the genre's most recognisable opening. Under `docs/ORIGINALITY.md` ("if a player would say that's from that other game, redesign it") it would need a strongly seafaring look: a sail-rigged balloon hull of tarred planks, net floats and rope.
- **B. Packet ship and lifeline mortar, as the GDD says.** A ship sails past offshore and players fire themselves inland on a mortar line, then steer a gull-kite. This is more distinctive and fits the rising-sea identity. It needs the launch to reach the far side of the island: the mortar sets the leap's speed and direction, and the route rule becomes "passes within X m of the coast" instead of "crosses land". About half a day of simulation work, with the same freefall and kite code.
- **C. Hybrid.** Keep the crossing route, but make the carrier a kite-towed sky raft launched from the packet ship. The GDD changes only in names.
**Recommendation:** B. It's the most original, and the GDD already committed to it. The current mechanics carry over: only the leap's start position and velocity change. If time before Monday is short, ship A's mechanics with C's names and look, and do B after v1.0.
Labelled `needs-eric`, so the scheduled build leaves this for Eric to decide.
claude
added this to the v0.6 Art, Audio and UI milestone 2026-11-11 17:27:33 +00:00
Done in aa79758: option B, as you chose. Sorry this sat for three days. The scheduled runs only checked needs-eric issues for a label change, not for your comment. I'll check for replies on every run from now on (noted in the DEVLOG).
What changed
The packet ship sails a seeded straight line past the island, offshore. It runs as close in as it can while keeping 45 m of open water from any land. Across 60 seeds it passes 15–140 m from the coast, and its route takes 27–59 s.
The lifeline mortar: jump fires you inland along the way you face. How high you aim sets the charge: level or lower is the least, and 45° or higher is the most. A level shot comes down a little inland; a full charge carries about 780 m on a plain fall, or about 990 m if you open the kite at the top of the arc. That's further than any land lies from any route on those 60 seeds (the farthest is about 907 m). The climb can't be steered, and the kite can't open until the arc tops out. Anyone still aboard at the end is fired towards the middle of the island.
The gull-kite replaces the wing-sail: salt-white sailcloth wings, the inner half raised and the outer half drooping like a gull's, with tarred tips, an ochre leading edge and a fanned tail.
Visuals: a two-masted packet with tanned square sails, a salt-white boot-top and a bowsprit, and the mortar on its timber bed, turned to face the island. A cord runs from the mortar to each player still climbing.
The bots, the soak and autoplay clients, and the tutorial pilot all aim using Drop.aim_charge(). The tutorial and PLAY.md explain the aim-to-charge rule.
Tested
make check: 783 tests pass, and so does smoke. New and changed tests cover route clearance and reach on 60 seeds, charge versus pitch, landings matching Drop.reach() both ways down, the climb, the forced shot at the end of the route, and the lifeline's ends.
make match (now seed 2): replay hash matches. make tutorial-check: 12/12. make client-check: UDP, 0 corrections.
make bot-matches (20) and make duo-matches (10) pass.
make balance (200): median match 13.7 min (was 13.3), and no weapon over 31%.
Captures rendered and looked at: vessels, match_ship, match_launch, match_glide, client_ship, client_glide, figure_states. The first launch capture showed the cord drawn wrong (scaled in world axes), which I fixed before pushing.
Not done or not verified
Sea deaths in make balance rose from 3.6% to 7.6%, still under the 10% gate. That's noted on H10 (#74).
Nothing has run on real GPU hardware, and how the launch feels is for your R6 playtest.
The v1.0.0 tag and draft release predate this change; see #64.
— tidewall-build
Done in aa79758: option B, as you chose. Sorry this sat for three days. The scheduled runs only checked `needs-eric` issues for a label change, not for your comment. I'll check for replies on every run from now on (noted in the DEVLOG).
**What changed**
- **The packet ship** sails a seeded straight line past the island, offshore. It runs as close in as it can while keeping 45 m of open water from any land. Across 60 seeds it passes 15–140 m from the coast, and its route takes 27–59 s.
- **The lifeline mortar:** jump fires you inland along the way you face. How high you aim sets the charge: level or lower is the least, and 45° or higher is the most. A level shot comes down a little inland; a full charge carries about 780 m on a plain fall, or about 990 m if you open the kite at the top of the arc. That's further than any land lies from any route on those 60 seeds (the farthest is about 907 m). The climb can't be steered, and the kite can't open until the arc tops out. Anyone still aboard at the end is fired towards the middle of the island.
- **The gull-kite** replaces the wing-sail: salt-white sailcloth wings, the inner half raised and the outer half drooping like a gull's, with tarred tips, an ochre leading edge and a fanned tail.
- **Visuals:** a two-masted packet with tanned square sails, a salt-white boot-top and a bowsprit, and the mortar on its timber bed, turned to face the island. A cord runs from the mortar to each player still climbing.
- The bots, the soak and autoplay clients, and the tutorial pilot all aim using `Drop.aim_charge()`. The tutorial and PLAY.md explain the aim-to-charge rule.
**Tested**
- `make check`: 783 tests pass, and so does smoke. New and changed tests cover route clearance and reach on 60 seeds, charge versus pitch, landings matching `Drop.reach()` both ways down, the climb, the forced shot at the end of the route, and the lifeline's ends.
- `make match` (now seed 2): replay hash matches. `make tutorial-check`: 12/12. `make client-check`: UDP, 0 corrections.
- `make bot-matches` (20) and `make duo-matches` (10) pass.
- `make balance` (200): median match 13.7 min (was 13.3), and no weapon over 31%.
- Captures rendered and looked at: `vessels`, `match_ship`, `match_launch`, `match_glide`, `client_ship`, `client_glide`, `figure_states`. The first launch capture showed the cord drawn wrong (scaled in world axes), which I fixed before pushing.
**Not done or not verified**
- Sea deaths in `make balance` rose from 3.6% to 7.6%, still under the 10% gate. That's noted on H10 (#74).
- Nothing has run on real GPU hardware, and how the launch *feels* is for your R6 playtest.
- The v1.0.0 tag and draft release predate this change; see #64.
— tidewall-build
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
Found while doing: W7 (#14)
The roadmap issue W7 calls the drop a "sky barge and wing-sail", and that's what is built: a carrier crosses the island at 150 m on a seeded straight route, and players leap, fall, open a sail and steer down. The GDD's core loop (written in F2) describes it differently: "A packet ship sails a straight line past the island. Each player picks a moment to leave by lifeline mortar and steers a gull-kite to the ground."
Only names and visuals depend on this. The simulation (
src/sim/drop.gd,Drop.NONE/ABOARD/FREEFALL/GLIDING) works for either. No client visuals exist yet; they come with the v0.6 art pass, so this needs deciding before then.Options
docs/ORIGINALITY.md("if a player would say that's from that other game, redesign it") it would need a strongly seafaring look: a sail-rigged balloon hull of tarred planks, net floats and rope.Recommendation: B. It's the most original, and the GDD already committed to it. The current mechanics carry over: only the leap's start position and velocity change. If time before Monday is short, ship A's mechanics with C's names and look, and do B after v1.0.
Labelled
needs-eric, so the scheduled build leaves this for Eric to decide.Yeah B is good. Let's do that.
Done in
aa79758: option B, as you chose. Sorry this sat for three days. The scheduled runs only checkedneeds-ericissues for a label change, not for your comment. I'll check for replies on every run from now on (noted in the DEVLOG).What changed
Drop.aim_charge(). The tutorial and PLAY.md explain the aim-to-charge rule.Tested
make check: 783 tests pass, and so does smoke. New and changed tests cover route clearance and reach on 60 seeds, charge versus pitch, landings matchingDrop.reach()both ways down, the climb, the forced shot at the end of the route, and the lifeline's ends.make match(now seed 2): replay hash matches.make tutorial-check: 12/12.make client-check: UDP, 0 corrections.make bot-matches(20) andmake duo-matches(10) pass.make balance(200): median match 13.7 min (was 13.3), and no weapon over 31%.vessels,match_ship,match_launch,match_glide,client_ship,client_glide,figure_states. The first launch capture showed the cord drawn wrong (scaled in world axes), which I fixed before pushing.Not done or not verified
make balancerose from 3.6% to 7.6%, still under the 10% gate. That's noted on H10 (#74).— tidewall-build