refactor: Remove static company sizes and KILL parallel arrays

[OH296]

---

Summary by cubic

Replaced static 501-slot arrays and parallel fields with per-unit properties and dynamic per-company rosters. Loops, serialization, and UI now respect actual roster sizes; ascension/gene-seed logic and ship occupant summaries are corrected.

• Refactors

  • Removed obj_ini.age/spe/god; use unit.age, unit.specials, unit.god_status across recruiting, moving, exports, Reclusiam, dialogue/events, AI, combat, and helpers.
  • Rosters resize to exact company size; scr_company_order rebuilds visible mirrors only (race/name/role/wep1/2/armour/gear/mobi).
  • find_company_open_slot grows per-company arrays and returns the new index; most loops now use array_length(obj_ini.TTRPG[c]) from 0.
  • Serialization iterates companies and each company’s TTRPG length (no 11×501 scan).
  • TTRPG_stats: age, specials, god_status are fields; roll_age sets age; marine_ascension is turn-based; armour/XP calculations read age.
  • Rewrote ship occupant listing to build pluralized role strings from a gathered unit list and vehicle roles.

• Bug Fixes

  • Reclusiam/jail and role counts read unit.god_status; UI toggles work with dynamic rosters.
  • Gene-seed recovery uses unit.age in after-combat and marine_ascension thresholds in kill-and-recover.
  • Detail panel shows rounded unit.age and ascension year.
  • Force counting, crusade, equipment checks, and exports iterate by per-company roster length.

^{Written for commit 44e9a25. Summary will update on new commits.}

[cubic-dev-ai]

All reported issues were addressed

<sub>Reply with feedback, questions, or to request a fix.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

<sub>Requires human review: Auto-approval blocked by 20 unresolved issues from previous reviews.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 4 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

All reported issues were addressed across 5 files (changes from recent commits).

<sub>Tip: Review your code locally with the cubic CLI to iterate faster.

Re-trigger cubic</sub>

[cubic-dev-ai]

0 issues found across 1 file (changes from recent commits).

Confidence score: 5/5

• Automated review surfaced no issues in the provided summaries.
• No files require special attention.

<sub>Requires human review: Auto-approval blocked by 17 unresolved issues from previous reviews.

Re-trigger cubic</sub>

[OH296]

/oc
Please, review this pull request:

Check for code quality issues of the newly added/changed code (docs/CODE_STYLE.md)
Look for potential bugs and regressions
Suggest improvements
Response style: you're the main maintainer of this codebase.

[opencode-agent]

Here's my review as the project maintainer.

---

Overview

This PR removes the fixed-size 501-slot arrays and parallel age/spe/god fields in favor of dynamic per-company rosters and per-unit properties. The direction is correct, but there are several critical bugs still unresolved and some code quality concerns.

---

P0 — Will crash at runtime

1. obj_ini/Create_0.gml:121 — race is not a 2D array, will crash on any company > 0 access

race = array_create(11, 501);

This creates [501, 501, 501, ...] — each element is the integer 501, not a sub-array. race[c][i] for c=1..10 indexes an integer and throws.

scr_initialize_custom at line 1482 only replaces race[0] with a real array via _init_marine_row(0, ...). Companies 1–10 are left as bare integers.

Fix: Same pattern as the other arrays — use array_create(11, []) and let scr_initialize_custom populate them, or revert to array_create_2d. But note issue #2 first.

2. obj_ini/Create_0.gml:123-137 — Shared array references for companies 1–10

name = array_create(11, []);
role = array_create(11, []);
// ... wep1, wep2, armour, gear, mobi, TTRPG

array_create(11, []) gives all 11 rows the same empty array reference. Companies 1–10 share a single array for name, role, TTRPG, etc. Writing to name[1][i] overwrites name[2][i] through name[10][i].

scr_initialize_custom only replaces company 0's array. Companies 1–10 remain aliased.

This is the exact bug cubic flagged three reviews ago and suggested array_create_ext for.

3. objects/obj_controller/Alarm_5.gml:293-296 — Null guard after property access

var _unit = fetch_unit([c,e]);
if (_unit.god_status == 10) {       // CRASH if _unit is undefined
    if (!is_struct(_unit)) {         // guard is too late
        continue;
    }

fetch_unit returns undefined for empty slots. The is_struct check needs to come before the property access.

4. scripts/scr_add_man/scr_add_man.gml:19-77 — Writes to undefined unit on normal recruitment path

Flow for other_gear == false:

1. Line 17: _company_slot = find_company_open_slot(...)
2. Line 19: scr_wipe_unit(...) — clears the slot to undefined
3. Line 20: _unit = fetch_unit(...) → undefined
4. Line 77: _unit.age = ... → CRASH on undefined
5. Line 116: _unit = new TTRPG_stats(...) — reassignment happens too late

The other_gear == true path is fine (switch blocks reassign _unit before line 77). The normal path is not.

5. scripts/scr_company_order/scr_company_order.gml:20 — Array index syntax error

function company_length(company){
    return obj_ini.TTRPG(company)
}

This calls TTRPG as a function, not indexing it as an array. Should be obj_ini.TTRPG[company].

This also means scr_ship_occupants at line 6 (for i <= company_length(co)) will crash on any ship occupant listing.

---

P1 — Will silently produce wrong results

6. scripts/scr_enemy_ai_d/scr_enemy_ai_d.gml:170 — Wrong property name

if (string_count("0", _unit.special) > 0) {

_unit.special (singular) doesn't exist. The field is specials (plural), used correctly everywhere else. The Warp Shadow psyker detection always fails as a result.

7. scripts/scr_random_event/scr_random_event.gml:712 — Same typo

if (_unit.role() == obj_ini.role[100][eROLE.CHAPTERMASTER] && string_count("0", _unit.special) > 0) {

Same _unit.special → _unit.specials fix needed.

8. scripts/scr_kill_unit/scr_kill_unit.gml:56-61 — Gene-seed recovery uses wrong time comparison

if (unit.marine_ascension > 30 && !obj_ini.zygote && !obj_ini.doomed) {
    obj_controller.gene_seed += 1;
}
if (unit.marine_ascension > 100 && !obj_ini.doomed) {
    obj_controller.gene_seed += 1;
}

marine_ascension is now obj_controller.turn at creation. A marine created on turn 5 is forever ineligible (5 > 30 is false). Previously this checked the calendar age (in years). The intent is an elapsed-time check, so these should use obj_controller.turn - unit.marine_ascension.

9. scripts/scr_marine_struct/scr_marine_struct.gml:259 — Turn number displayed as calendar year

marine_ascension = obj_controller.turn;

Previously this stored millenium * 1000 + year (a calendar year like 41735). Now it stores a turn count like 101. The detail panel at scr_unit_detail_text.gml:52 renders "{round(age)} years old. Ascended to an Astartes in the {ascension_date} year." — which will display "101 year". Either the display needs converting, or the field needs both values.

---

P2 — Design & code quality

10. scripts/scr_initialize_custom/scr_initialize_custom.gml:1482-1489 — Only initializes company 0

_init_marine_row(0, ...) and TTRPG[0] = array_create(501) only set up company 0. Companies 1–10 are left with bare integers (race) or shared arrays (everything else). The deserialize at Create_0.gml:303-306 also hardcodes for _mar <= 500 with TTRPG[_coy][_mar] — which will crash on integer race and corrupt the shared arrays.

This needs to be consistent — either initialize all companies at startup or ensure the dynamic-resize path is universally reliable.

11. scripts/scr_company_order/scr_company_order.gml:122-131 — Compaction resizes arrays but other consumers don't know

The company ordering function truncates arrays to _new_length (the number of actual marines). Other consumers like scr_max_marine, scr_check_equip, scr_crusade previously iterated to 501 and may now index out of bounds. Need to verify they all use array_length() dynamically.

12. scripts/scr_chapter_managent_events/scr_chapter_managent_events.gml:276 — No struct guard

for (var i = 0; i < array_length(obj_ini.TTRPG[0]); i++) {
    scr_move_unit_info(pick.company, 0, pick.marine_number, i);
    break;

Missing is_struct(obj_ini.TTRPG[0][i]) before the move. Also this loop always breaks after the first iteration — is that intentional?

---

Summary

Must fix before merge:

1. race = array_create(11, 501) → crash on any company > 0
2. Shared [] references for name/role/wep1/wep2/armour/gear/mobi/TTRPG → data corruption across companies 1–10
3. _unit.god_status accessed before is_struct guard in Alarm_5.gml
4. _unit.age assignment on undefined in scr_add_man normal path
5. TTRPG(company) instead of TTRPG[company] in company_length → crashes scr_ship_occupants

Should fix:
6-7. _unit.special → _unit.specials in two files
8-9. Gene-seed elapsed-time check, marine_ascension display units

This is a substantial refactor with a lot of moving parts. The shared-array initialization and the race 1D-vs-2D issue are the most concerning because they corrupt data silently or crash immediately on any interaction with companies 1–10.

github run