Vault

zappatic MOO1 RE stream — Part 2 notes

Per-part extraction (auto-caption transcript → structured notes). Values are as heard; hex is often caption-garbled — confidence marked per item. Digest: zappatic-re-stream.

Source: part02-3tdKnTGFq2I.txt — https://www.youtube.com/watch?v=3tdKnTGFq2I

Source: Vault/raw/videos/moo1-re/part02-3tdKnTGFq2I.txt (auto-captions). Numbers below are as heard; captions garble hex, so treat any hex value as low-confidence unless marked otherwise. “unknownNN” is the streamer’s own field-index label inside his struct definitions, not a byte offset.

Summary

• Mapped the planet’s per-turn build accumulators: ship-building progress (unknown19), defense/base progress (unknown45), factory/industry progress (unknown20) [0:05:00–0:15:00].
• Identified planet fields: explored flag (59), unrest/rebellion state (87?), stargate (47), shield class (49), owner/”colonized by” (24), and two planet-modifier fields (hostile/fertile/gaia) [0:20:00–1:20:00].
• Reworked the tech-research struct: the “target” is really base cost, the progress fields were off by one, and he found a per-domain flag that forces or blocks the next breakthrough, plus a per-domain “count researched” field [1:20:00–2:00:00].
• Diffed savegames before and after researching Deep Space Scanner. Found the researched-tech list: 6 arrays × 60 uint8 tech IDs (string indexes) inside the per-player block, previously mislabelled “player ship configuration” [2:05:00–2:50:00].
• Used memory value-search to find the empire reserve (32-bit) and several tech-screen label values. Most of the latter are display-only [2:55:00–3:10:00].
• Gave himself tech IDs (antimatter drive, thorium cells) to test ship range and contact. Unlimited range triggers contact with all races [3:15:00–3:50:00].
• Started a races struct with per-opponent spy sliders, spy jobs, security slider and spy counts. Ran a symbol-overlap check to place it just before tech research [3:30:00–4:10:00].
• “Diplomat gone” turned out to be screen-only state, not saved. His write breakpoint failed to fire, which he blamed on a bug in his own tool. Unresolved at end [4:00:00–4:15:00].

Tooling & setup

• Own memory-viewer / editor app: the window scales now; it lists weapons, specials and all research names, typed in off-stream [0:00:00]. Fields can be shown as checkboxes, enum drop-downs, signed/unsigned 8/16/32-bit and arrays. Arrays of enums lose the drop-down because he never implemented that [3:40:00].
• Symbol file = JSON [1:15:00]. It holds top-level symbols with segment:offset, e.g. a planets array typed planet[108] (108 planets, as heard), and struct definitions with ordered fields. These are applied to live emulator memory. He edits it in VS Code and hot-reloads symbols [2:55:00, 3:35:00]. He likens it to the original C/C++ struct Planet [1:15:00].
• Struct-size discipline: he keeps a fixed struct size when splitting fields (per-player block “should be 575 bytes”, which turned out to be 576) so later repeated blocks stay aligned [2:35:00–2:45:00].
• “Find symbol overlaps” tool: finds structs whose ranges share physical addresses. He uses it to place a new struct and to check whether a found address is already mapped [3:30:00, 4:05:00].
• Memory value search (Cheat-Engine style: search value, change it in game, “next search” to narrow) inside “our range for Orion” [2:55:00, 3:50:00]. Also searches for byte sequences in a segment [2:20:00].
• Own bindiff tool for comparing two save slots side by side [0:50:00, 2:05:00]. His savegame editor also writes saves. He checked its on-disk writes with a separate hex editor because it is “not tested thoroughly” [2:15:00].
• Write breakpoints in his debugger. They failed to fire on a 16-bit and on a byte value, which he called his own bug, to fix later [4:10:00–4:15:00].
• Compiler: he remarks “this is Borland’s C++” [1:10:00]. Context unclear (it may answer a chat question). Confidence: low-med.
• The game manual, in text form and as the official MicroProse PDF (tech screen, p.31), was used to interpret UI [1:30:00–1:35:00, 3:25:00].
• Ghidra and disassembly were not used in this session. It was all live-memory and savegame work.
• Addresses heard (garbled, low confidence): reserve around “4244” [2:55:00]; spy slider at “423F”, races struct started at segment offset “…CC” / “4230” [3:30:00–3:35:00]; diplomat-gone screen flag at “47AD”/”478AD” [4:00:00]; tech amounts around “B33435” [2:00:00].

Method

• Play-and-poke loop: pick an unknown field, compare its value across planets/states, write values (often “big” or recognisable ones) through the editor, then check (a) the UI, (b) whether it persists or resets next turn, and (c) the effect after End Turn. If a field resets every turn, it is probably derived or recomputed (e.g. cost recalculated, labels refreshed) [0:00:00–0:30:00, 1:20:00].
• Arithmetic confirmation: predict the next value as old + production and check it, e.g. 342 + 106 = 448 vs observed 449 (“one off”) [0:05:00–0:10:00], and 183 + 52 = 235 exact [0:10:00]. Then force it, e.g. ship progress 1580 vs cost 1590 gives a ship next turn [0:10:00].
• Enum discovery by sweep: write 0..N and record which values the game accepts or reverts. Shield accepts 0/5/10/15/20, not 25 [0:35:00]; unrest 0/1/3 [0:30:00]; modifier values spill into unrelated strings past the end of the table [1:15:00].
• Bitfield vs enum test: unrest with 5 and 15 (all low bits set) did not trigger, so it is a plain value, not a bit check [0:25:00–0:30:00].
• Savegame diffing: save in slot 1 just before the event (colonize, research completes) and in slot 2 just after, then bindiff [0:50:00, 2:05:00]. To isolate which changed byte matters, he duplicates the “before” save and patches candidate bytes one at a time, reloading each time [2:10:00–2:20:00]. He copied a whole block over to rule a region out [2:15:00].
• Gotcha, count-gated lists: patching an ID into an array shows nothing unless the matching “count” field is also raised, because the UI loops only N entries [2:10:00, 2:30:00].
• Locating save data in memory: search for a unique neighbour, such as the ship name “Polaris”, instead of non-unique byte runs, then count over to the target [2:20:00]. The reverse also works: take bytes around a memory value and search for them in the save to find where it lives in the file [3:30:00].
• Plant a recognisable value to find a field in the save. When the planted value never appears in the save, the field is transient UI state [4:10:00].
• Off-by-one recovery: he spotted that his “progress” labels were shifted by one because an extra field sits between the per-domain values, and renumbered them [1:40:00–1:45:00].
• Dead ends: lowering ship cost does nothing because cost is recomputed each turn [0:05:00, 3:20:00]. Several planet unknowns (14, 16, 18, 21, 22, 77) showed no visible effect [0:20:00, 1:10:00–1:20:00]. Contact flags were lost because a search wasn’t narrowed before a new game reset them [3:55:00].

Findings — data structures & file formats

Planet struct (live planets[108]; also in the save, first section)

• Save planet record order as seen in diff: name (12 chars, garbage after the terminator), x, y (e.g. 144, 92), star colour (yellow=0), size (small=0), “flicker timer” counter, … [0:50:00]. med
• unknown19 = ship-building progress (accumulated production; compared to ship cost) [0:10:00]. high
• unknown45 = defense/base building progress, next to the base count [0:10:00]. high
• unknown20 = industry/factory building progress (carries the fractional factory) [0:15:00]. med-high
• unknown59 = explored boolean, meaning a ship has passed. Setting it reveals planet details [0:20:00, 1:05:00]. high
• unknown87 (as heard) = unrest state enum: 0 none, 1 unrest, 3 rebellion. Not a bitfield [0:25:00–0:30:00]. med (index heard once)
• unknown47 = stargate present, boolean 0/1 [0:30:00]. high
• unknown49 = planetary shield class enum: 0, 5, 10, 15, 20 (25 invalid). It is cleared next turn unless research or other state backs it, so another field or progress is involved [0:35:00]. med
• unknown24 = owner / “colonized by”, signed 16-bit: −1 (65535) uncolonized, 0 = player, 1..5 = opponent slots. Opponent races are random per game, so this is not a fixed race enum [0:55:00–1:00:00]. Setting it to 0 hands you the planet [1:00:00]. high
• unknown10 tracks population, possibly max-pop related. Setting current population to 0 crashed, perhaps a divide-by-zero (his guess) [1:00:00–1:05:00]. low
• Two planet-modifier fields combine into the label under the planet [1:10:00–1:15:00]. Modifier “1”: 0 hostile?, 1 empty, 2 fertile, 3 gaia, 4 empty/corrupt, 5 spills into an unrelated string. The other field holds values such as hostile and Forex Tech (hostile = −1 or 0; Forex Tech is the last entry of a preceding string array) [1:15:00]. low-med, mapping partly muddled in speech
• Unknowns with no visible effect: 14 (2/3/4 across planets), 16 (0/1), 18, 21, 22, 77 (−1 on some planets), plus a set of six zero fields that reset every turn [0:20:00–0:45:00, 1:10:00–1:20:00]. Earlier fiddling with them may have dropped a planet’s population to 1 [0:40:00].

Per-player block (in the save after the planets; repeated once per player, then trailing unknown data) [0:45:00–0:50:00, 2:50:00]

• Contents: 6 ship designs, new ship names, unknown data, unlocked research arrays, ship counts array. He renamed it from “player ship configuration” to player info [2:50:00]. high
• Unlocked research: 6 arrays (computers, construction, force fields, planetology, propulsion, weapons) × 60 uint8 slots. Each entry is an index into that domain’s tech-name string array. The array is authoritative: putting 45 into computers made Battle Computer Mk X selectable in ship design [2:20:00–2:45:00]. high
• Only the first count entries are valid; the rest is uninitialized garbage [2:30:00]. The game sorts the array by ID when a new tech completes, and next-research offers depend on the highest ID held [2:50:00]. high
• Block size ended at 576 bytes (he expected 575) [2:45:00]. med
• Ship designs: size field values small=2, medium=15, large=100, huge=700 alongside costs 8 / 49 / 270 / “5090”. This is not a cost input [0:00:00]. A design field forced back to 0, an unknown flickering field auto-corrects to 2, and unknown27 is not the engine [0:00:00, 3:15:00]. low
• Ship counts per design, e.g. 2 scouts + 1 colony ship [2:40:00–2:45:00]. med

Tech-research struct (memory; per player; repeated per player in the save) [1:20:00–2:00:00, 2:05:00]

• Per domain: next-research ID (“target” choice), base cost (the manual’s term; he first called it target), progress, and an in-between flag (renumbered unknown56/57) [1:35:00–1:45:00]. med-high
• In-between flag: 0 shows %, 1 shows the bulb, 3 something lower. Setting it to 1 can force the breakthrough next turn even at 952/18000, but it sometimes blocks completion until cleared [1:25:00, 1:45:00–1:50:00]. low-med
• Starting around unknown91: count of techs researched per domain (e.g. computers 11, construction 3). This is the array length for the unlocked-research arrays. Setting construction to 10 shows a generic “Construction Technology” entry [1:55:00–2:00:00]. high
• Diff on Deep Space Scanner: computers count 1→2, base cost changed (“406” hex heard, about 1600), next target 4→10 (0x0A), progress reset to 1, a “level” 1→4, 3→5, 0x39→0x3A, 0x50→0x51 [2:05:00]. low for the hex
• unknown13+14 = reserve (BC), 32-bit, verified by exceeding 65535 [2:55:00–3:00:00]. high
• unknown11+12 = total-income label, 32-bit and recomputed (display only) [3:00:00]. med
• Display labels: unknown29 computer scan range (persists), unknown34 robot control, unknown38 propulsion ship range (label only; it does not change travel), unknown37 factory waste stored as value ×10, unknown23 terraform label, waste-reduction label just before it [3:00:00–3:10:00]. med
• unknown15 = reserve slider position, range 0–200 (not 0–100) [3:10:00]. med
• Ordering learned from the overlap check: races block → tech research (starts with bytes “B0 16B”, as heard) → planet orbiting ships [4:05:00]. med

Races / espionage struct (new, partial) [3:30:00–4:10:00]

• Per-opponent spy sliders (u16; first found by value search; 5 opponents) at about unknown21+ [3:35:00]. med
• Spy jobs enum around unknown39/40: 0 hide, 1 sabotage?, 2 espionage? (he swapped the order once) [3:40:00]. low-med on 1 vs 2
• unknown44 = overall security slider, 0–200 scale [3:40:00]. med
• unknown46 = spy count per opponent [3:45:00]. med
• Five suspicious values of 2, one per opponent, meaning unknown [3:50:00]. The struct ends at field 47 to avoid overlapping tech research [4:10:00].
• “Diplomat gone” flag (heard 47AD/478AD) is UI-screen state, not saved. A planted value never reached the save [4:10:00]. med

Findings — game mechanics & formulas

• Build progress accumulates as production per turn into per-planet accumulators. A ship is produced when ship progress ≥ ship cost. Lowering cost alone does nothing because cost is recomputed from the design each turn [0:05:00–0:10:00, 3:20:00]. high
• Factories can build at fractional rates (0.2 per year means one every 5 turns), with the remainder carried in the industry accumulator [0:15:00]. med
• Population has no hidden accumulator; growth is applied straight to current population [0:15:00]. low-med
• Research points are global, summed from all planets and split by the slider ratios directly into each domain’s progress [0:15:00–0:20:00]. med
• After base cost is met, completion is probabilistic and the % rises with overspend, but it is not simple: 1704/1280 gave 46%, about 9408/~6000 gave 34%. The % appears even slightly below base cost (base 12500: 12500→5%, 12600→6%, 11300→1%, threshold around 11261/11262) [1:25:00–1:35:00]. Formula unknown. low
• Progress decays when a domain gets 0% allocation (computers dropped about 700 from 6713) [1:45:00]. med
• On completion, progress resets to 1 and the overshoot is lost [1:25:00]. med
• Base cost is recomputed from the researched set; inserting advanced techs changed the next cost to 75 [2:50:00]. low-med
• Starting techs come free (e.g. nuclear missile and laser for weapons; battle scanner for computers) [1:55:00, 2:45:00]. med
• Ship travel range is limited to N parsecs from the nearest colony, set by fuel-cell tech (manual: reserves allow 5 parsecs; thorium cells = unlimited) [3:20:00–3:25:00]. Getting unlimited range immediately triggered contact with all races [3:30:00, 3:50:00]. med
• The labels on the tech screen (scan range, ship range, factory waste, terraform +N) are derived values, and editing them does not change gameplay [3:00:00–3:10:00]. med
• Tech IDs heard: computers 1 battle scanner, 4 deep space scanner, 10 (0x0A) Battle Computer Mk II, 45 (0x2D) Battle Computer Mk X, plus ECM Jammer Mk 5 and “advanced computer tech 8”. Construction 1 reserve fuel tanks, 3 improved industrial tech 9, 10 “duroid” (duralloy) armor. Propulsion 41 thorium cells [2:25:00–2:35:00, 3:25:00]. med (5 was also heard as “Mark 2 battle computer”, which conflicts with 10; low)

Named code/data

All are field labels in his JSON symbol file, not code addresses.

• Planet: ship building progress (unk19), defense building progress (unk45), industry building progress (unk20), explored (unk59, checkbox), unrest enum (unk87?), stargate (unk47), shield enum planet shield class (unk49), colonized by (unk24, enum planet colonizers), planet modifier 1/2 [0:10:00–1:15:00].
• Player block: unlocked research arrays of struct achieved research <domain> (u8 enum) × 60 per domain; ship counts [2:30:00–2:50:00].
• Tech research: base cost (renamed from target), progress per domain, “achieve next turn" flags, `amount of researched` (from unk91), `reserve` (u32, unk13+14), total income planets (u32), computer scan range (29), computer robot control (34), propulsion ship range (38), construction factory waste (37, /10), planetology terraform (23), reserve slider (15) [1:50:00–3:10:00].
• Races: spy slider[opponent], spy jobs (enum hide/sabotage/espionage), overall security slider (44), spy count (46), diplomat gone (UI, around 47AD) [3:35:00–4:05:00].

Open questions / unresolved

• Formula for the research completion % and the exact role of the per-domain “achieve/block” flag [1:30:00–1:50:00].
• The red dot on the tech screen; the manual does not explain it [1:30:00–1:35:00].
• Planet unknowns 14, 16, 18, 21, 22, 77 and six resetting zero fields; terraforming and shield progress fields need the matching research first [1:20:00].
• Which field actually grants a planetary shield [0:35:00–0:40:00].
• Trade, treaties, war/alliance, relation (“amiable/uneasy”) and contact flags were not found [3:00:00, 3:50:00–4:00:00].
• The five values of 2 per opponent and the reverting fields after the ship counts [2:50:00–2:55:00, 3:50:00].
• Ship-design unknowns (the size→2/15/100/700 field, the auto-correcting one); why some researched engines and thorium cells don’t appear in design lists [0:00:00, 3:15:00–3:30:00].
• Write breakpoint not firing, attributed to his tool; persistent diplomat state still unknown [4:15:00].
• The big block between planets and players in the save, which changes every turn [3:30:00].