Oasis Wildlife Simulator

01 · The brief

Twenty-seven animals, one month at a time

Zone food is what GetZoneFood() returns for each zone. Reptiles don’t eat zone food in the model, and Scrub Country isn’t listed in GetZoneFood(), so it gets the 300 default.

02 · Class design

One base class, three families, six species

double KeepInRange(double value) {    if (value < 0.0) {        return 0.0;    }    else if (value > 100.0) {        return 100.0;    }    else {        return value;    }}

03 · The monthly loop

One month, in order

One_Month() updates every animal in a fixed order, then runs a death pass. Order matters here: CheckHabitat() and CheckTemperature() set values outright instead of adding to them, so running them last would overwrite the changes made before them.

1. 01

   Habitat

   • CheckHabitat()
   • CheckTemperature()
   • CheckNestingSite()

   CheckHabitat() and CheckTemperature() set comfort (and, for reptiles, energy and hydration) outright, so they run first. Birds’ CheckNestingSite() also runs first, but only adjusts comfort.

2. 02

   Food, season & species

   • CheckFoodAvailability()
   • FeedingBehaviour()
   • ShedSkin()
   • Migrate()
   • CheckFeeding()
   • CheckBurrowingAbility()
   • …

   Adds to and takes from the stats: food against need, the season’s effects and each species’ own rules. Each mammal gets its zone’s food divided by the total number of mammals; birds run their hunting or foraging check first, because it sets how much they catch.

3. 03

   Death pass

   • CheckLife()
   • BasicDeathCheck()
   • erase()

   Runs after every animal has been updated. The species’ own rule comes first, then the shared one. The dead are erased, and their destructors correct the counts.

bool BasicDeathCheck(Animal& a) {    if (a.getHP() <= 0.0) {// HP at or below 0 means dead        return true;    }    else if (a.getEnergyLvl() <= 0.0) {// Energy at or below 0 also means dead        return true;    }    else if (a.getHungerLvl() >= 100.0) {// if hunger is 100 or higher then dead (starved)        return true;    }    else {        return false;// anything in between is alive    }}

// KOALASfor (int i = 0; i < Koala_count; ) {    bool dead = false;// start by assuming this is alive    //species-specific first (calls Koala::CheckLife on this object)    if (!koalas[i].CheckLife()) {        dead = true;    }    else if (BasicDeathCheck(koalas[i])) {// then the general rules (a Koala IS an Animal)        dead = true;    }     if (dead) {// if this died this month        d.koala = d.koala + 1;// record 1 death in the Deaths struct return to main()        koalas.erase(koalas.begin() + i);// remove koala number i, the rest slide up one    }    else {// still alive        i = i + 1;// move to the next index only when we didn't delete    }}

04 · Debug log

Fourteen bugs, each traced to a cause

Every fix is written up in a comment where it was made: what went wrong, why, and what changed. The log below follows those comments, symptom first, because that’s how each bug showed up.

1. BUG-01Animal_count fell 27 → 0 → −27 (every other counter by its own population on each call), and the tables stopped printing rows after the first one.main.cpp

   CausePrintProperties() took the six vectors by value. The copies were made by the compiler’s own copy constructor, which never runs Animal(), so the counters never went up for them. But ~Animal(), ~Mammal() and ~Koala() still ran when the copies were thrown away, so every call took each counter down by the number of animals it counts.

   FixPass the vectors by reference (vector<Koala>&). The same fix meant One_Month() finally changed the real animals instead of copies.

2. BUG-02Species counts went negative after deaths.main.cpp

   CauseThe death loop decremented Koala_count by hand, and erase() destroyed the object, which ran the destructors that decrement it again. Every death was counted twice.

   FixDeleted the manual decrements. Each count now goes up in a constructor and down in a destructor, and nowhere else.

3. BUG-03Every koala and wombat “died” in April while eating perfectly.main.cpp · Animal.cpp

   CauseNothing limited the stats. A well-fed mammal’s hunger kept falling (−270 after a year), and a “hunger ≤ −100 is dead” rule was killing the best-fed animals. Lizard HP reached 235 “percent”.

   FixKeepInRange() clamps every setter to 0–100, and that rule is gone. The death checks now work cleanly because HP or energy can land exactly on 0.

4. BUG-04Every kookaburra and emu was starving, every single month.main.cpp

   CauseTheir daily food needs (400–480 for kookaburras, 550–700 for emus) were more than their zones held: 200 at Lagoon Landing for the kookaburras, 500 at Coral Cove for the emus.

   FixA bird now eats what it catches: zone food × hunting or foraging efficiency. Food needs were rescaled to that scale, in the same order as before.

5. BUG-05Kookaburras got +40 comfort every month, wherever they lived.main.cpp

   CauseCheckNestingSite() was passed the bird’s own nesting preference, so it compared the preference with itself. And once it was compared with the zone’s habitat, "Wetlands" would never have equalled "wetlands".

   FixIt now gets the zone’s real habitat from GetZoneHabitat(), with lower-case habitat words throughout. A kookaburra that wants grassland but lives in the wetlands is now uncomfortable, as it should be.

6. BUG-06Over two or three years, every kookaburra slowly got too weak to hunt and starved.Species.cpp

   CauseThe seasonal multipliers (×0.9 in winter, ×1.1 in summer) were saved back into the stat each month, so they compounded, and integer maths rounded down every time.

   FixFrilled lizards, kookaburras and emus now keep a “normal” value. Each month the season adjusts a copy of it for that month only.

7. BUG-07Animals with no energy left didn’t die.Species.cpp

   CauseThe check was getEnergyLvl() == 0.0. Energy moves in steps like −15, so it can jump from 10 to −5 and never be exactly zero.

   FixgetEnergyLvl() <= 0.0.

8. BUG-08Weak but well-fed kookaburras died, and starving ones lived.Species.cpp

   CauseThe hunger test in Kookaburra::CheckLife() was backwards (≤ 60), even though a higher HungerLvl means hungrier everywhere else.

   Fix> 60, consistent with the mammals’ > 70.

9. BUG-09Every kookaburra ran out of energy by its second month.AnimalFamilies.cpp

   CauseBirds start on 50 energy and migrating costs 40 a month in a migratory season, but nothing ever gave any back. ActivityCycle() was optional in the brief, so it had been left empty.

   FixActivityCycle() now models roosting: +20 energy a month, and +10 more for a bird that’s comfortable in its nest.

10. BUG-10Latent undefined behaviour in the death loop.AnimalFamilies.cpp

    CauseMammal() left its bool, int and double fields uninitialised until main() seeded them. erase() copies animals, and copying an uninitialised bool is undefined behaviour. Because main() happened to seed everything first it never fired, but nothing guaranteed that.

    FixMammal() now sets defaults, the same way Reptile() and Bird() already did.

11. BUG-11A mammal’s comfort ignored whether it was living in its preferred habitat.main.cpp

    CauseCheckHabitat() was written but never called. It also sets comfort outright (= 100) instead of adding to it, so calling it last would wipe out the month’s other comfort changes.

    FixIt’s called first in each mammal’s monthly update, and the same rule applies to CheckTemperature() for reptiles. Order matters.

12. BUG-12Table rows reshuffled after a death, so one animal couldn’t be followed from month to month.main.cpp

    CauseDead animals were removed by copying the last animal into their slot and calling pop_back().

    Fixerase() keeps everyone in order, and the loop index only moves on when nothing was removed.

13. BUG-13Bad input quietly ran a simulation nobody asked for.main.cpp

    CauseInvalid entries were forced to the nearest valid value (0 months became 1, month 15 became December), and typing letters jammed cin for every read after it.

    FixThe program re-prompts until the input is valid, using cin.clear() and cin.ignore() to recover from a failed read.

14. BUG-14The program wouldn’t link: “undefined reference to EBTLizard::CheckLife()”, plus “undefined symbol” errors for counters such as Koala_count and Kookaburra_count.Species.cpp

    CauseEach one was declared in a header but never defined anywhere.

    FixOne definition per symbol in its .cpp file, with extern in the header. That’s the rule every global counter in the program follows.

// (same "normal value" idea as CheckThreatResponse above)void Kookaburra::CheckHuntingSuccess(string season) {    int hEff = normalHuntingEfficiency;     if (season == "Winter") {        hEff = (hEff * 90) / 100;    }    else if (season == "Summer") {        hEff = (hEff * 110) / 100;    }     if (hEff > 100) {// it's a percentage, so it can't go past 100%        hEff = 100;    }     huntingEfficiency = hEff;

05 · Run it

The real program, every start month

$ ./oasis
Hello Dr. Elara, please enter how many months you would like to simulate: 24

Next, please enter the month you want to start this simulation:
[ menu: 1 - January … 12 - December ]
Enter your choice (1-12): 1
[ initial values + 24 monthly tables ]

 ################################# Simulation Summary ###################################
 Months simulated: 24
 Animals still alive: 21 of 27
 Total deaths: 6
0 Koala Deaths in total (5 left)
0 Wombat Deaths in total (7 left)
1 Eastern Blue-Tongued Lizard Deaths in total (2 left)
1 Frilled Lizard Deaths in total (2 left)
3 Kookaburra Deaths in total (2 left)
1 Emu Deaths in total (3 left)

06 · Ten years in

Every start month ends the same way

07 · Next steps

Where it could go next

1. N1Let well-fed animals recover health.Nothing gives a koala HP back, so the model guarantees every koala dies. A small gain for a good feed would let it settle into a real balance.
2. N2Add births.The brief marked BirthProcess() as “omitted”, so it was left empty and the population can only fall. With births, a long run would mean something.
3. N3One loop instead of six.Make CheckLife() and a monthly update virtual in Animal and keep every animal in one vector<unique_ptr<Animal>>. Then adding a species no longer means editing main().
4. N4Rules as data, with tests.Put zone food, season multipliers and death thresholds in tables, with unit tests on the edge cases behind bugs 03, 07 and 08.