news

有趣的模拟进化的程序 由国外一生物学家开发 十分有趣

[I] = Internal code change only; no extral effects should be present.
[O] = Change to Output or Viewer (no changes to functionality)
[F] = New Feature available for use, or feature removed (inactive by default)
[B] = Bug fix; something that used to work wrong will now be correct.
[M] = Major change; avida will function differently in certain circumstances.


For version 2.0 beta 6:

September, 03
[B] Widened number display fields in CPU viewer so that data can be read
    correctly.

[F] Added buttons to select number format (binary, decimal, or
    hexadecimal) in Register, Inputs, and Stacks views of CPU viewer in
    GUI.

[B] Minor bugfixes in 4-stack cpu parasite code.

[F] Added option to count phenotypes based on tasks completed.

For version 2.0 beta 5:

August, 03
[B] Workaround for bug in Qt/Windows: setInsideSpacing can't be safely
    called, so removed those calls.

[F] Addition of CPU stats to instruction viewer in GUI.

[B] Organism inputs are no longer per-cell, but per-organism; resolved
    related problem of request of now nonexistent inputs.

[B] Restored missing initializations of subclass of cPopulationInterface
    class in qt-viewer.

[B] Added documentation for some existing events.  Changed some
    formatting of events.html documentation so file looks better with
    Safari web browser.

[B] Added missing files to visual studio project.

[B] Restored compatibility with older versions of gcc.

[F] Added parasite tracking for 4-stack cpus, added events
    "print_dom_parasite" and "detail_parasite_pop", added default
    organsisms and settings for the 4-stack cpu.

[F] Added Bounded Grid/Torus (world topology) switch to population.
    Leaving torus as the default.

[B] Fixed some problems in mutation mapping in analyze mode.

July, 03
[B] Text viewer Zoom Screen Mini-Map fixed; Modified Map screen now in
    color; Thread Map screen now shows number of threads used by each
    organism.

[F] Completion of 4 Stack Hardware.  Inject command and parasite
    abilities now implemented.  Text Viewer environment view mode in
    progress.

[I] Added integer version WriteBlockElement to file output routines.

[B] Correction of minor logical error in reaction handling.

For version 2.0 beta 3:

June, 03
[F] Avida now runs and builds on Windows.  Many minor changes were made for this, 
    .pro files were added to be used with qmake, the way the configure script searches
    for the ncurses library was modified, pdcurses was incorporated instead of ncurses,
    trio was added to fix some string problems, and some very minor code changes were
    implemented to allow for some differences in the visual studio compiler. 

[F] new 4 stack hardware is implemented but not yet fully functional.  It's compatible
    with primitive and ncurses modes, however will cause qt-viewer to crash if used. 
    No functionality associated with the original hardware has been changed.

[B] set up constant time slicer to skip over empty cells, this had been throwing an assert
    error in debug mode.

[B] removed Visual Studio.net support files as they are out of date and no longer work properly.

May, 03
[F] introduced new 4 Stack CPU model.  Removed 3 registers in CPU and replaced with
    4 stacks. Added 4th NOP which allows use of all 4 stacks.  Also changed complement system 
    for nops.  Pairs are now as follows: A,C; B,D; C,A; D,B;

[B] newer C++ compilers want scoping on templates defined inside class definition, event
    if they are only used inside the class.

[B] fixed problem with incorrect copying of instruction libraries when one instruction set was
    initialized to another one.

[F] set up Avida to pring out a genesis file if none exists.

[I] working on setting up Avida to generate an instruction set file in none exists.  Attached 
    60 character or less descriptions to each of the default instructions, and attached a flag to all 
    instructions declaring they are default.

[I] added default event files in C++ that are generated from perl so Avida can be compiled in Windows
    without running perl script by hand as long as no new events are created or modified.

[B] updated .pro files so don't have to change any properties in Visual Studio to compile

[B] The hardware factory has always been creating CPU's of type cHardwareCPU even if everything else
    was assuming cHardware4Stack.  This all works properly now.

[I] added pdcurses.h and curses.lib to Avida so one doesn't have to actually check out the whole 
    pdcurses suite to run on Windows.

[I] added third party suite "Public Domain Curses" for Win32 compatability.

[F] added event print_genotype_map that prints a Matlab file of the genotype IDs to a default file
    "genotype_map.m"

[B] added empty directory aclocal to CVS so don't have to create it by hand every time build Avida
    from scratch

[I] update .pri files for Windows builds

[F] added Visual Studio.net support files for ncurses viewer, primitive mode, and qt-viewer.

[I] cHardware4Stack now hooked into Avida and working properly

[I] added new hardware type cHardware4Stack.  Not hooked into rest of Avida yet...

[I] re-engineered cFile class to work based off of arrays instead of linked lists.  Each line is 
    now stored as part of a structure which has other info associated with it, such as line numbers
    for better error reporting.  Not being used yet, but available...

[I] added HARDWARE_TYPE to genesis file and setup config to load it into Avida, it's not being used yet.

[I] modified test_cup.cc to improve the default inputs in the test CPUs so they are more similar
    to those in the environment, and ideally give more accurate answers.
    Serup the default instruction set to be hooked in with a proper instruction library so that
    they can be used outside of a population (i.e. in analyze mode)

[B] fixed bug in testCPU

[I] changed cInstLib to cInstSet and cInstSuperLib to cInstLib.  Renamed inst_lib.{cc,hh} to 
    inst_set.{cc,hh} and inst_superlib.hh to inst_lib.hh.  Updated other files accordingly to 
    reflect this change

[I] included instruction which injects a parasite into it's own thread.  Also changed count.dat
    so that info regarding parasites is appended to the end of the file.  Removed previous parasite
    info which is now obsolete.

For version 2.0 beta 1:
[B] edited Makefile.am to reflect the fact that built event source file depends are build-tree relative
    but cpu source file depends are source-tree relative

[I] removed unused sourcecode from beneath qt subdirectory

[I] introduced cInstSuperLib to contain a complete library of available instructions corresponding to
    a given hardware class.  cInstLib now indexes into arrays provided by cInstSuperLib

[B] mkdir() is not defined in win32 so added win32_mkdir_hack.hh to define it as a macro so can
    use in windows version

[B] modified namespaces, adding 'using namespace std' appropriately throughout Avida to allow
    compilation on Windows

[I] set up qt builds to be performed using qmake on seperately maintained .pro files

[B] fixed error in new tasks, now cast int to double before taking log

[I] added code to allow spatial resources

[I] allow long lines in an input file to be split across multiple physical lines by use of
    a \ character at the end of the line as a continuation marker

[F] added two routines to aid the printing of spatial resource data:
    writeblock - allows for the output of individual grid points
    writeraw - allows any line to be written anywhere in an output file

April, 03
[F] added event kill_rectangle which allows user to kill all organisms in a rectangle
    described by points (X1, Y1) and (X2, Y2).

[B] no longer crashes when population dies out.

[I] added version of Max function with type double.  added function Neighbor
    that works with population and spatial resource arrays to find the current neighboring
    cells (even at edge of matrix).

[B] fixed time slicer problem where drastic drop in maximum merit would slow down a run 
    considerably.  Added warning when analyze mode tries to load in a string sequence
    thats supposed to represent a genome, but maximum char goes beyond instruction set size.

[I] previous runs can no longer be replicated even when a random seed used is known
    due to extensive reorganization of portions of code that handle divides.

[F] initial implementation of sex.  Now exists divide-sex and divide-asex commands,
    former facilitating recombination between two offspring by swapping a region of 
    their code.

[B] fixed hardware getting incorrect info as to if it lived through birth process.

[B] removed mutation rate drifting options and old style crossovers.

[I] assembled framework for sexual divides.  Mostly involved pushing some of the 
    final offspring creation to the population and out of the organism.

For version 2.0 beta 0:
[I][O][F][B][M]
    Avida has been almost entirely rewritten, and this NEWS file is
    woefully out of date.  In an upcoming beta this NEWS file will be
    brought up to date.
    
For version 1.6.1:
[B] Fixed a problem with CHILD_SIZE_RANGE in the genesis file -- it used to
    only compare the child size to the *new* parent size; now it will properly
    compare it to the original parent size (and make sure that the new parent
    size also falls into the same legal range)

[O] Added an Average Fitness output in the primitive viewer.

[O] Added a column 10 to the stats.dat output file that indicates the
    depth of the most recent coalescent genotype.

[I] Removed "defs.hh" dependency from assert; stay out of debug move by
    default -- only enter it when DEBUG is defined.

[I] Rebuilt data_file.??, simplifying the internal structure and using a
    proper linked list.  Also removed dependency on defs.hh.

[I] Removed dependencies in string.?? ; fixed some bugs in Compare()
    that didn't turn up before, but seem real.  Often '\0' was being
    saught when a cString was being compared to a char *, when the cString
    data does not end in a \0 (a seperate size measure is kept)

[I] Removed global memory object.  It was used throughout the code, but
    most allocations did *not* make use of it.  We should use some kind
    of solid memory tracking method, but this one was less than optimal.

[I] In the viewer, shifted the maps from being normal arrays to using the
    tArray object; also kept them all local to maps (rather than being
    allocated in one place and freed in another like they were)

[I] Shifted environment connection lists such that they are now part of
    the environment from the beginning and only initialize (not created)
    by the population.  Plus some general cleanup and organization in them.

[I] Moved all cList objectes over to the cleaner, templated tList.  This
    allowed for some standardization in our current way of doing things,
    a cleaner interface (and simplification of some code).  In the future
    we may want to move all of this over to the STL list template, but for
    the moment this was much easier.

[I] Cut down on the number of includes from tools.hh; most of them are
    not required.

[I] Converted all of the "TRUE" and "FALSE" occurances to "true" and "false"
    throughout the entire code.  Changed int to bool in most appropriate
    cases.

[I] Moved code_array.?? entirely over to assert from debug.

[I] Anihilated old geneology stuff.