feat(biology): fund offspring from a distinct inherited genome
Foundation / checks (ecs) (push) Successful in 2m27s
Foundation / checks (deploy) (push) Successful in 4m15s
Foundation / checks (network) (push) Successful in 21m4s
Foundation / checks (client) (push) Successful in 22m22s
Foundation / checks (tsan) (push) Successful in 31m40s
Foundation / postgres-tools (push) Canceled after 0s
Foundation / server-image (push) Canceled after 0s
Foundation / image (push) Canceled after 0s
Foundation / checks (postgres) (push) Canceled after 45m38s
Foundation / checks (checks) (push) Canceled after 45m43s

This commit is contained in:
dingo
2026-10-19 13:09:52 -04:00
parent 936ce36b3f
commit 5d8aa4d2e3
11 changed files with 394 additions and 4 deletions
+13
View File
@@ -1,5 +1,18 @@
# Project status and next steps
## Distinct-genome offspring funding - 2026-10-19
[ADR-0164](adr/0164-distinct-genome-offspring-funding.md) implements the first
biology operation for generational structural inheritance. An unchanged parent
can fund a child with a distinct inherited anatomy and starting demand, with
exact energy/element closure and no free tissue. Siblings retain the selected
plan while archive records keep the parent's expressed genome separate.
Both full MSVC/GCC Release suites pass **83/83**. Focused ASan/UBSan, clang-tidy
and formatting pass; [validation](engineering/VALIDATION.md) records exact scope.
Stored offspring plans, mutation timing, birth control handoff, retained NPC
behavior and choice UI remain next work. Current live/default biology is unchanged.
## Gameplay alignment: structural inheritance and succession - 2026-10-19
The maintainer accepts [ADR-0163](adr/0163-structural-inheritance-and-offspring-succession.md):
@@ -0,0 +1,76 @@
# ADR-0164: Fund offspring from a distinct inherited genome
## Status and scope
Accepted for the first portable biology implementation segment of ADR-0163,
2026-10-19. Extends ADR-0032/0125/0137. This is a complete pure funding proposal,
not admission of new owner rules, stored heritable state or player succession.
## Decision
Add `fund_offspring(parent, offspring, growth, reserve, temperature, share)` and
`offspring_cost(parent, offspring, temperature, share)` to the existing biology
budding API. Both genomes are immutable validated records. Require equal schema
and development versions in this first adapter. Refuse a reused genome ID with
different contents. Archive ancestry, authority, placement and durable publication
remain the caller's responsibility; a cost proposal grants none of those rights.
The expressed parent determines reproductive capability, grown/undamaged
eligibility, and the existing reproduction investment, including construction
heat and any admitted reproduction-skill coefficient. Divide that investment
channelwise by `share` exactly as existing twins do; remainders stay with the parent.
The offspring genome determines its own body, birth growth and energy provision.
Select the largest growth in 1..1000 whose body fits every investment matter
channel. Remaining invested matter is child reserve. The parent additionally
funds 30 seconds of that child's own full demand at its birth growth. This uses
the existing trial provisioning constant, not an adult-body or survival grant.
As before, refuse if the largest body's full provision is unaffordable; do not
silently select a smaller child to reduce that energy bill. A non-reproducing
offspring is permitted if the reproductive parent can construct it. A parent
without reproduction cannot use a reproductive child's capabilities to bud.
Cross-schema/development reproduction is explicitly unsupported pending a later
versioned conversion contract, rather than mixing construction laws implicitly.
Every channel closes exactly:
`parent reserve before = parent reserve after + child body + child reserve + heat`
Heat has energy only and body has matter only. The parent body is never spent or
rebuilt. Inputs remain unchanged, and refusals return no partial transition.
Cost uses the same funding path against a bounded ample reserve; exact shown
cost funds the proposal and one-unit shortage in any required channel refuses.
The shared investment and fixed tissue composition bound all supported genomes
well below that ample value. Unsupported or impossible proposals have no cost.
Existing `fund_bud` and `bud_cost` delegate with the same genome in both roles.
All old signatures, numerical behavior, rule meanings and golden bytes remain.
No dependency, library boundary or existing codec changes.
## Alternatives, benefits, costs and risks
Rebuilding the parent into the selected child form violates ADR-0163. Pricing
the entire proposal from the child would let it supply reproductive capability
or change the parent's already expressed investment. Copying the old parent body
into a differently shaped child can fail elemental composition and geometry.
Separating the two derivations supplies an actual funded biological operation
without introducing a second balance or arbitrary mutation fee.
The cost is two explicit genome inputs and an incompatible-genome refusal.
This adapter retains existing reproduction/rounding limitations; it does not
select final balance, new morphology, heredity timing or a future hybrid species
model. Funding a possible newborn does not prove mature ecological viability.
## Qualification and next boundaries
Check old same-genome equality across supported founders, distinct membrane/size
and demand variants, maximal feasible growth, all-channel closure and shortages,
twins, sterile children, parent eligibility, incompatible versions and reused IDs.
Retain real archive parent/child links through a codec round-trip without changing
the parent's expressed genome. Tests run under NDEBUG, analysis and ASan/UBSan.
Next define and implement versioned expressed/heritable owner state, operator
expression timing and retained plan history, then atomic birth/control choice,
bounded persistent NPC behavior and readable birth UI. Current default/DEV rules
continue exact-copy budding; SQL admission, census, migration and old snapshots
do not acquire these features by accepting this pure operation.
+2
View File
@@ -1,5 +1,7 @@
# Architecture decision records
- [ADR-0164: Distinct-genome offspring funding](0164-distinct-genome-offspring-funding.md) - pure funded reproduction separates the expressed parent's investment from the inherited child's anatomy and provision; old same-genome behavior retained, owner/control integration pending.
- [ADR-0163: Structural inheritance and one-way offspring succession](0163-structural-inheritance-and-offspring-succession.md) - maintainer-accepted target gameplay: offspring structural traits, irreversible birth-time control choice and NPC-parent behavior/reproductive continuity; implementation pending.
- [ADR-0162: Bounded unscripted wildlife generations](0162-bounded-unscripted-wildlife-generations.md) - a predeclared finite player-origin raw/cohort measurement with production controller, hold-outs and controls; no biological or deployment change.
+16
View File
@@ -22,6 +22,22 @@ be implemented before this behavior is offered. Choice timing and pause policy
remain explicit implementation decisions. The M11 decisions and diagram below
describe the old baseline, not a prohibition on the accepted target behavior.
## Implemented portable offspring funding
[ADR-0164](../adr/0164-distinct-genome-offspring-funding.md) implements
`fund_offspring` and `offspring_cost` in the biology budding API. The expressed
parent supplies reproductive eligibility and its own investment; a distinct
inherited genome supplies the child's body, largest feasible birth growth and
starting demand. Every elemental channel and energy close exactly. The parent
is not rebuilt. Changed schema/development laws and conflicting same-ID genomes
refuse. Two share-funded children cannot manufacture the parent's invested matter.
Tests cover changed anatomy and demand, sterile children, exact cost/shortages,
twins and retained archive identities. Existing same-genome helpers delegate to
this operation with identical parent/child inputs. This is a pure proposal;
stored heritable state, mutation expression timing, succession, NPC behavior and
birth UI remain unimplemented. Current live rules still produce exact-copy buds.
## Historical M11 baseline and open strategies
Locked: irreversible death, independent descendants and no offline progression.
+5
View File
@@ -12,6 +12,11 @@ control and census state do not implement that target. These require coordinated
versioned biology, authority, recovery and UI work; earlier exact-copy generation
evidence does not close that gap or establish the intended generational gameplay.
[ADR-0164](../adr/0164-distinct-genome-offspring-funding.md) now implements the
pure biological funding step with distinct parent/offspring genomes and exact
closure. Existing same-genome behavior remains. Persistent heritable plans,
expression timing, authority handoff and NPC/UI integration are still required.
## Persistence boundary resolved by 12.3b
ADR-0142 adds dedicated profile 5, population trace 1 and schema 44; raw profile 4
+32
View File
@@ -1,5 +1,37 @@
# Validation report
## 2026-10-19 - Distinct-genome offspring funding
[ADR-0164](../adr/0164-distinct-genome-offspring-funding.md) adds a complete pure
funding proposal with distinct expressed-parent and inherited-child genomes. The
parent pays its own admitted investment; child anatomy, largest feasible birth
growth and provision come from the child's genome. Old helpers delegate with
the same genome in both roles. No owner/save/rules, default opening, SQL or live
gameplay changes; stored plans and control succession are still pending.
Executed locally:
- MSVC 19.44 Release full CTest **83/83 passed, 20.65 s**, with NDEBUG.
- GCC 14.2 Release full CTest **83/83 passed, 15.06 s**, with NDEBUG.
- Clang 19.1 ASan/UBSan budding, distinct-genome offspring and funded maturation
tests **3/3 passed, 30.76 s**, without sanitizer reports.
- clang-tidy 19 on implementation and tests, repository clang-format 19 and diff
whitespace checks pass. Initial test analysis caught an optional dereference
hidden behind a test macro; an explicit runtime guard resolves it. Refusal
checks guard `expected::error()` under NDEBUG as well.
Tests cover distinct membrane/size and movement-demand plans, largest funded
growth, exact cost and each channel shortage, twin investment, sterile offspring,
parent capability/growth/damage, impossible anatomy, schema/development mismatch,
same-ID conflict, sibling plan round-trip and retained archive parent/child
identity. All old tests/golden meanings remain. No private database rerun: the new
mixed-genome path is not admitted by the existing cellular owner or SQL bundle.
The required DEV window in this push expires **2026-10-19 21:09:13 UTC**.
Rendered policy passes: 75 objects in 74 files, 138 rendered objects in 17 overlays,
zero failures. The unchanged policy self-test matrix was not repeated. No live
deployment or World0 operation was performed.
## 2026-10-19 - Accepted structural inheritance and birth succession direction
[ADR-0163](../adr/0163-structural-inheritance-and-offspring-succession.md) records