Command line interfaced ogame-conflict simulator. v1.5.7: - legal: - License project under Apache License v2.0. v1.5.6: - Bugfix-Prod: - Bad value of CDR corrected. v1.5.5: - Feature-Prod: - the genetic engine can now include the true pilled ressources (after destruction of some attacking ships and re-computation of capacity). - Bugfix-Prod: - Display, in perl mode, the metal-metal lost from ships etc. for each ressources. v1.5.4: - Bugfix-Dev: - minor bug in heap/genetic algo. - Bugfix-Prod: - correct recycling stat output in -o p mode. - Feature-Prod: - add a -w option to divide the scoring by the flight time / wave time. v1.5.3: - Bugfix-Dev: - Race condition due to a shared pointer in os_fleet_t ctx->fleet. - Bugfix-Dev: - Fix format for long/long long. v1.5.2: - Bugfix-Prod: - Refactor threadpool to avoid deadlock. v1.5.1: - Bugfix-Prod: - Remove a bug introduced in 0.9.7, no-recycling was confused with recycling mode. - Feature-Prod: - Now in perl output don't display not-rentable fleet. v1.5.0: - Feature-Prod: Adding SMP support. - Bugfix-Prod: Fix Knuth fast rand to work on 64bit. - Cosmetic-Dev: Improving indentation. ;) v1.0.0: - v0.9.9 v0.9.9: - Bugfix-Prod: - deut_consumed was wrong in perl output. - Feature-Prod: - Remove invested ressources if we are in no-invest mode. v0.9.8: - Feature-Prod: - Now no more architecture dependant, compile as well on x86_64 as 686. - Bugfix-Prod: - Fix --with-apr behavior, now use --with-apr-config - Bugfix-Prod: - Fix typo that was underestimating the memory. (MemFree instead of memfree). - Feature-Dev: - Flush stdout/stderr. - Feature-Dev: - use rand() instead of Knuth fast rand() on 64 bit arch, because it fails... - Feature-Dev: - Set max individual to 1024. - Feature-Dev: - improve genetic fitness by removing some float arithmetic. - Bugfix-Dev: - Fix signed/unsigned bugs. v0.9.7: - Feature-Prod: - Add switches investment/loss/recycling. - Feature-Prod: - Add output switches. v0.9.6: - Feature-Dev: - Fix autoconf/Makefiles to handle apr-config. - Bugfix-Prod: - fix traq speed change. v0.9.5: - Feature-Prod: - Add 0.76 ogame features. v0.9.4: - Bugfix-Prod: - Fix Moon parsing when resources < 0. - Feature-Prod: - Improve first fleet tries in guess mode to test only powerfull ships. - Feature-Prod: Modify mutation function to help the decreasing of ships. v0.9.3: - Bugfix-Prod: - flight time in guess mode used incorrect technos values. v0.9.2: - Bugfix-Prod: - Remove recycling from script mode. v0.9.1: - Fixed time limit for genetic algorithm. - Experimental: Fix flight time (theoricaly working, not praticaly). - Add no missile and best defense mode. - Bugfix: read meminfo on 2.4.x kernel. v0.9.0: - Script version featuring: - 3 selectable guess mode: no-buffer fleet, full fleet, 'script'. - Inactivity timeout for genetic algorithm, if no new best individual has been found for xx seconds, stop the generations. - Genetic algorithm best individual timestamped. - script mode does not allow: loss (at least for the 16 simulations it does), investment, missiles or buffer ships, it does stop by default at 30 seconds of genetic algorithm inactivity. It will display a short summary easier to parse for a perl script. Scoring is the number of ressources won. v0.8.6: - Bug apparently fixed. v0.8.5: - Try to cast unsigned long long operation because a bug appeared... v0.8.4: - Now use unsigned long long (64 bits) to store various informations that are overflowed in case of big simulation (> 200k ships). v0.8.3: - Now use Buffers and Cached from /proc/meminfo to look for free mem. v0.8.2: - Change memory usage according to what is available on the system. v0.8.1: - Bugfix in MIP generation, don't add MIP if there's 0 def. v0.8.0: - Add MIT and MIP calculation, so input has changed see -h to check. - Change ratio between (editable in os_conf.h) ressources to 1.0 1.0 1.0. - refactor no-buffer implementation (improve speed). - Add shell scripts to make the user simplify its life :P v0.7.1: - If stdin is incorrect output, don't segfault... - Add Porte de Saut & Base Lunaire to Batiments. - Handle low def ennemy (take in account fleet capacity) & Refactor no-buffer. v0.7.0: - Add the -d stdin option. That read a spy report (limitation : this spy report must contain last line : "Probabilité de destruction de la flotte d'espionnage"). - Guess mode : generate a first individual with full - user fleet. - Guess mode : take deut consumption in account (as a deut loss). v0.6.1: - Spy report parsing. (not interfaced yet, but working regexp). v0.6.0: - Guess mode: Remove deathstar when in no-buffer mode. Take in account the existing fleet to guess the cheapest one. Increase number of simulation by fleed tried (from 4 to 16). Add ressource spent to get the guessed result. v0.5.0: - Bugfix in rapid fire, clo was resisting too much to the edlm. - Simulation mode: Add Debris, losses statistics. - Guess mode: Change fitness function, fix debris calculation. v0.4.3: - Huge optimization by defining my_rand instead of rand(). - Optimisation in mutation function. v0.4.2: - Improve performance of shootship. v0.4.1: - Improve performance of rapidfire. - Tweak ratio of ressource (I'll make it configurable). v0.4.0: - Add no-buffer / remove weak ship from simulation into guess mode. - genetic algorithm, select different mating partenair than those who already have been written on. - add mutation mode that does not affect other ship. - Bugfix in ship shoot. v0.3.1: - Change ratio metal/cristal/deut to something closer to the rate we usually harvest from inactives. - Include number of round in the fitness (the lesser the better). - Improve fleet reduction (less rand() call) if random choose too many ships. - Change mutation, the removal of ship of a type implicate an addition of same price in a ship of different type. Mutation now is able to reset randomly rarely one fleet too. v0.3.0: - Add optimizations, featuring a new mode to guess the best attacker/defender in front of a specific fleet. v0.2.1: - Add optimizations, change compilation option to improve efficiency on 686. - Featuring best/worst case now. v0.2: - Basicly working, optimised, but no bench against speedsim have been done yet. - Featuring average case. v0.1: - Non working proof of concept (optimizations that didn't work).