PhyG.cabal

Cabal-Version: 3.12
Name:          PhyG
Version:       1.2
Stability:     Alpha
Build-Type:    Simple
Tested-With:
  GHC == 9.2.8
  GHC == 9.4.7
  GHC == 9.6.3
  GHC == 9.8.1
  GHC == 9.10.1
  GHC == 9.12.1

Author:        Ward Wheeler <wheeler@amnh.org>
Copyright:     © 2015 Ward Wheeler and The American Museum of Natural History
License:       BSD-3-Clause
License-File:  doc/LICENSE

Maintainer:    Ward Wheeler <wheeler@amnh.org>
Homepage:      https://github.com/AMNH/PhyG#readme
Bug-Reports:   https://github.com/AMNH/PhyG/issues

Synopsis:      Performs heuristic search of phylogenetic graph space via scoring abstract input data.
Description:
  Phylogenetic Graph (PhyG) is a multi-platform program designed to produce phylogenetic graphs
  from input data and graphs via heuristic searching of general phylogenetic graph space. The
  bio-informatics framework libraries of the broader Phylogenetic Haskell Analytic Network Engine
  (PHANE) project are the foundation upon which PhyG is constructed. PhyG offers vast functionality,
  including the optimization of unaligned sequences, and the ability to implement search strategies
  such as random addition sequence, swapping, and tree fusing. Furthermore, PhyG can generate
  outputs in the form of implied alignments and graphical representations of cladograms and graphs.
  What sets PhyG apart from other phylogenetic analysis programs, is the extension to broader
  classes of input data and phylogenetic graphs. The phylogenetic graph inputs and outputs of PhyG
  include trees, as well as softwired and hardwired networks.

Data-Files:
  doc/Authors.md
  doc/Funding.md
  doc/LICENSE

Extra-Doc-Files:
  doc/Authors.md
  doc/Changelog.md
  doc/Funding.md
  doc/LICENSE
  README.md

Extra-Source-Files:
  doc/Authors.md
  doc/Funding.md
  doc/LICENSE
  ffi/external-direct-optimization/*.h


Flag Enforce-Sanity
   Description: Ensure that all warnings are handled
   Default:     False
   Manual:      True

Flag Enforce-Timestamp
   Description: Disable multi-threading
   Default:     True
   Manual:      False

Flag Forego-Sanity
   Description: Disable extensive and helpful compiler warnings
   Default:     False
   Manual:      True

Flag Single-Threaded
   Description: Disable multi-threading
   Default:     False
   Manual:      True

Flag Super-Optimization
   Description: Apply extremely agressive and extremely time consuming optimization passes
   Default:     False
   Manual:      True

Flag Use-LLVM
   Description: Utilize the LLVM code generator backend
   Default:     False
   Manual:      True

Flag Memoize-Via-ConcurrentHashtable
   Description: Memoized hash-table uses the "Concurrent Hashtable" package's implementation
   Default:     True
   Manual:      True

Flag Memoize-Via-IORef
   Description: Memoized hash-table uses the "IORef" implementation
   Default:     False
   Manual:      True

Flag Memoize-Via-ManualLock
   Description: Memoized hash-table uses the "ManualLock" implementation
   Default:     False
   Manual:      True

Flag Memoize-Via-ReadWriteLock
   Description: Memoized hash-table uses the "ReadWriteLock" implementation
   Default:     False
   Manual:      True

Flag Memoize-Via-Semaphore
   Description: Memoized hash-table uses the "Semaphore" implementation
   Default:     False
   Manual:      True

Flag Memoize-Via-TVar
   Description: Memoized hash-table uses the "TVar" implementation
   Default:     False
   Manual:      True


-- Global deviations from Haskell98
Common ffi-build-info

--  If !os(darwin)
--    CC-Options:
--      -rdynamic --std=c11
--
--  Else
--    CC-Options:
--      --std=c11

  CC-Options:
    --std=c11
    -- this for LINUX compile
    -fPIC

  -- Here we list all directories that contain C & C++ header files that the FFI
  -- tools will need to locate when preprocessing the C files. Without listing
  -- the directories containing the C header files here, the FFI preprocessor
  -- (hsc2hs, c2hs, etc.) will fail to locate the requisite files. Note also,
  -- that the parent directory of the nessicary C & C++ header files must be
  -- specified. The preprocessor will not recursively look in subdirectories for
  -- header files!
  Include-Dirs:
    ffi/external-direct-optimization

  -- Specify the header files as required source files here.
  -- Do not specify them in the c-sources or cxx-sources stanzas.
  -- This is required for sdist and install commands to work correctly.
  Includes:
    alignCharacters.h
    alignmentMatrices.h
    c_alignment_interface.h
    c_code_alloc_setup.h
    costMatrix.h
    debug_constants.h
    dyn_character.h
    ukkCheckPoint.h
    ukkCommon.h

  C-Sources:
    ffi/external-direct-optimization/alignCharacters.c
    ffi/external-direct-optimization/alignmentMatrices.c
    ffi/external-direct-optimization/c_alignment_interface.c
    ffi/external-direct-optimization/c_code_alloc_setup.c
    ffi/external-direct-optimization/costMatrix.c
    ffi/external-direct-optimization/dyn_character.c
    ffi/external-direct-optimization/ukkCheckPoint.c
    ffi/external-direct-optimization/ukkCommon.c

-- This is required as compile option to make static binaries on Linux at least
-- need to haev llvm-(verions) tools installed to link properly
-- enable-executable-static -fpic -fPIC (multiple times--no idea why but including both cases seems to mater on 
-- commandline)--sometimes have to run twice or more to get link correct (with differnet case for PIC)
-- seems inconsistent with 9.12.1

-- Global build directives
Common build-defaults

  -- If we want to enforce that the compile-time timestamp is accurate,
  -- then we must ensure that we recompile the appropriate module(s).
  -- This sets the CPP flag so that the modules can query if they should
  -- always be recompield and then ask GHC to recompile them.
  If flag(Enforce-Timestamp)
    CPP-Options:
      -DENFORCE_TIMESTAMP

  -- Select the memoized hash-table implmentation to be used!
  If flag(Memoize-Via-ReadWriteLock)
    CPP-Options:
      -DMemoize_Via_ReadWriteLock

  Elif flag(Memoize-Via-IORef)
    CPP-Options:
      -DMemoize_Via_IORef
            
  Elif flag(Memoize-Via-ManualLock)
    CPP-Options:
      -DMemoize_Via_ManualLock
            
  Elif flag(Memoize-Via-Semaphore)
    CPP-Options:
      -DMemoize_Via_Semaphore
            
  Elif flag(Memoize-Via-TVar)
    CPP-Options:
      -DMemoize_Via_TVar

  -- The flag of the default implementation goes last to give others preference.
  Elif flag(Memoize-Via-ConcurrentHashtable)
    CPP-Options:
      -DMemoize_Via_ConcurrentHashtable

  -- If no flags are set, meaning that the default flag was manually disabled,
  -- then proceed with the default implementation anyways!
  Else
    CPP-Options:
      -DMemoize_Via_ConcurrentHashtable

  -- We want these Language Extensions enabled in *all* modules by default.      
  Default-Extensions:
    BangPatterns
    DerivingStrategies
    ImportQualifiedPost
    LambdaCase
    NoGeneralizedNewtypeDeriving
    RoleAnnotations
    UnicodeSyntax

  Default-Language:
    GHC2021

  If flag(Enforce-Sanity)
    GHC-Options:
      -Werror

  If !flag(Forego-Sanity)
    GHC-Options:
      -- Sanity check warnings
      --   1. Include all warnings by default
      -Weverything
      --   2. Exclude the undesirable warnings
      -Wno-all-missed-specialisations
      -Wno-implicit-prelude
      -Wno-inferred-safe-imports
      -Wno-missing-import-lists
      -Wno-missing-kind-signatures
      -Wno-missing-safe-haskell-mode
      -Wno-monomorphism-restriction
      -Wno-redundant-bang-patterns
      -Wno-type-defaults
      -Wno-unsafe
      -- Remove these exclusions later
      -Wno-missed-specialisations
      -Wno-x-partial

  If os(darwin)
    Ld-Options:
      -Wl,-no_fixup_chains


-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
--
-- Preamble of package description and reusable definitions from above has ended
-- and below we list all build targets of the package. Build targets include:
--
--   * Primary executable(s) for public usage
--
--   * Exposed sub-libraries for public consumption as program dependancies
--
--   * Benchmarks for executables and sub-libraries
--
--   * Test-suites for executables and sub-libraries
--
--   * Additional executables for non-public, diagnostic purposes
--
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --


-- Phylogenetic Graphs
--
-- This is the main program of this package, supporting fully featured
-- phylogenetic trees, networks, and forests.
Executable phyg

  Import:
    build-defaults,
    ffi-build-info,

  Main-Is:
    phygraph.hs

  GHC-Options:
--     -feager-blackholing --  may be causing parallel loop errors
    -O2
    -rtsopts
    -Wall
    --  Maybe this is required for good paralellism?          
    -fno-omit-yields 
    -- There is/are known compiler bug(s) with this optimization in GHC 9.6.2 (and earlier?).
    -- Hence we explicitly disable it and hope that it removes segfaults on some architechtures.
    -- -fno-polymorphic-specialisation

--  Ghc-Prof-Options: 

  If flag(Use-LLVM)
    GHC-Options:
      -fllvm

  -- Setup the Run-Time System (RTS) defaults
  -- Note:
  --   * Usage of -AL8 was rejected  should be >= A
  --   * usage of --nonmoving-gc causes large memory blow-ups on cluster

  -- If the the program should only use a single thread,
  -- Then use the following RTS options:
  If flag(Single-Threaded)
    GHC-Options:
      "-with-rtsopts=-A64m -AL128M -H1024m -n4m"

  -- Otherwise, enable threading
  -- and additionally set the maximum number of threads as the default:
  Else
    GHC-Options:
      -threaded
      "-with-rtsopts=-N -A64m -AL128M -H1024m -n4m -qa -qm"

  -- If need for a high performance executable was specified,
  -- then at compile-time GHC will allocate additional time and memory
  -- to explore and apply the most agrressive optimization possible.
  If flag(Super-Optimization)
    GHC-Options:
      -fexcess-precision
      -fexpose-all-unfoldings
      -flate-specialise
      -fmax-simplifier-iterations=16
      -foptimal-applicative-do
      -fspec-constr-count=8
      -fspec-constr-keen
      -fspecialize-aggressively
      -fstatic-argument-transformation
      -fstg-cse
      -fstrictness-before=1
      -fstrictness-before=2
      -fstrictness-before=3
      -fstrictness-before=4
      -funbox-small-strict-fields
      -funbox-strict-fields

  Build-Depends:
    PHANE-alphabet,
    PHANE-dynamic-character,
    PHANE-dynamic-character-element,
    PHANE-evaluation,
    PhyG:dynamic-character,
    PhyG:tcm,
    PHANE-PhyloLib,
    base >= 4.10,
    bimap,
    bv,
    bv-little,
    bytestring,
    containers >= 0.6,
    cpuinfo,
    deepseq,
    directory,
    encode-string,
    fingertree,
    file-embed,
    filepath >= 1.4.100,
    fingertree,
    fgl >=5.8,
    foldl,
    gitrev,
    graphviz   >=2999.20,
    hashable,
    inflist,
    logfloat,     
    MissingH >= 1.4.3,
    MonadRandom >= 0.6 && < 1.0,
    mmark,
    modern-uri,
    optparse-applicative,
    parallel,
    prettyprinter,
    process,
    random,
    random-shuffle,
    split,
    template-haskell,
    text >= 2.0.1 && < 3.0,
    text-builder-linear >= 0.1.1.1,
    text-short,
    th-lift-instances,
    time,
    unbounded-delays,     
    unliftio-core >= 0.2 && < 1.0,
    vector >= 0.13 && < 1.0,
    zlib
    -- currently doesn't compile but could be useful
    -- lots of good functions
    -- Graphalyze,

  -- include the source files needed in src for github library
  -- this for local library files
  HS-Source-Dirs:
    app
    src

  Other-Extensions:
    CPP
    TemplateHaskell

  Autogen-Modules:
    PackageInfo_PhyG
    Paths_PhyG

  Other-Modules:
    CommandLineOptions
    CommandLineOptions.Display
    CommandLineOptions.Types
    Commands.CommandExecution
    Commands.CommandUtilities
    Commands.ProcessCommands
    Commands.Transform
    Commands.Verify
    Complexity.CodeStrings
    Complexity.Constants
    Complexity.Graphs
    Complexity.Huffman
    Complexity.Utilities
    Debug.Debug
    GraphOptimization.Medians
    GraphOptimization.PostOrderSoftWiredFunctions
    GraphOptimization.PostOrderSoftWiredFunctionsNew
    GraphOptimization.PreOrderFunctions
    GraphOptimization.Traversals
    Graphs.GraphOperations
    Input.BitPack
    Input.DataTransformation
    Input.FastAC
    Input.ReadInputFiles
    Input.Reorganize
    Input.TNTUtilities
    PackageInfo_PhyG
    Paths_PhyG
    Reconciliation.Adams
    Reconciliation.Eun
    Reconciliation.ReconcileGraphs
    Search.Build
    Search.DistanceMethods
    Search.DistanceWagner
    Search.Fuse
    Search.GeneticAlgorithm
    Search.NetworkAddDelete
    Search.Refinement
    Search.Search
    -- Search.Swap
    Search.SwapMaster
    Search.SwapV2
    Search.WagnerBuild
    Software.Credits
    Software.License
    Software.Metadata
    Software.Metadata.Embedded
    Software.Metadata.TimeStamp
    Software.Preamble
    Software.SplashImage
    Support.Support
    System.Timing
    Types.DistanceTypes
    Types.Types
    Utilities.DistanceUtilities
    Utilities.Distances
    Utilities.LocalGraph
    Utilities.LocalSequence
    Utilities.TcmHash
    Utilities.ThreeWayFunctions
    Utilities.Utilities
    -- GraphOptimization.PostOrderFunctions


-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
--
-- Collection of sub-libraries which both, compose the package's primary program
-- (PCG/PhyG), and are also exposed for external consumption by other programs.
--
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --


-- Library for performing string alignment on dynamic characters.

-- Provides various metrics for scoring static characters and
-- performing string alignment on dynamic characters.

Library dynamic-character

  Import:
    build-defaults,
    ffi-build-info

  HS-Source-Dirs:
    lib/dynamic-character/src

  Build-Depends:
--    PHANE-alphabet,
    PHANE-dynamic-character,
    PHANE-dynamic-character-element,
    PhyG:tcm,
    PhyG:utility,
    base                     >= 4.11      && < 5.0,
--    bv-little,
    containers               >= 0.6.2     && < 1.0,
    matrices                 >= 0.5       && < 1.0,
    monad-loops              >= 0.4       && < 1.0,
    primitive                >= 0.7.1     && < 1.0,
    vector                   >= 0.12.0.3  && < 1.0,

  Exposed-Modules:
--    Bio.DynamicCharacter
--    Bio.DynamicCharacter.Measure
--    Bio.DynamicCharacter.HandleGaps
    DirectOptimization.Pairwise
    DirectOptimization.Pairwise.Visualization
    DirectOptimization.Pairwise.Swapping
    DirectOptimization.Pairwise.Ukkonen
    DirectOptimization.PreOrder

  Other-Modules:
    DirectOptimization.Pairwise.Direction
    DirectOptimization.Pairwise.Huge
    DirectOptimization.Pairwise.Internal
    DirectOptimization.Pairwise.Slim
    DirectOptimization.Pairwise.Slim.FFI
    DirectOptimization.Pairwise.Wide


-- Library for working with TCMs and SCMs in various representations.

-- General purpose library for working with transition cost matrices (TCMs)
-- and symbol change matrices (SCMs). Specialization options are provided
-- for the discrete metric (non-additive) and the L1 norm (additive) TCMs &
-- SCMs. Exposes a memoized binding for sparsely indexed, large TCMs.

Library tcm

  Import:
    build-defaults,
    ffi-build-info

  HS-Source-Dirs:
    lib/tcm/src

  Build-Depends:
--    PHANE-dynamic-character,
    PHANE-dynamic-character-element,
    PhyG:utility,
    binary                   >= 0.8       && < 1.0,
    base                     >= 4.11      && < 5.0,
    containers               >= 0.6.2     && < 1.0,
    concurrent-extra         >= 0.7.0     && < 1.0,
    concurrent-hashtable     >= 0.1.8     && < 1.0,     
    deepseq                  >= 1.4       && < 2.0,
    hashable                 >= 1.3       && < 2.0,
    hashtables               >= 1.2       && < 2.0,
    QuickCheck               >= 2.14      && < 3.0,
    mono-traversable         >= 1.0       && < 2.0,
    semigroupoids            >= 5.3       && < 7.0,
    stm,
    vector                   >= 0.12.0.3  && < 1.0,
    vector-binary-instances  >= 0.2.5     && < 1.0,

  Exposed-Modules:
    Data.Hashable.Memoize
    Data.Hashable.Memoize.ViaConcurrentHashtable
--    Data.Hashable.Memoize.ViaConcurrentHashtableOpt
    Data.Hashable.Memoize.ViaIORef
    Data.Hashable.Memoize.ViaManualLock
    Data.Hashable.Memoize.ViaReadWriteLock
    Data.Hashable.Memoize.ViaSemaphore
    Data.Hashable.Memoize.ViaTVar
    Data.MetricRepresentation
    Data.TCM
    Data.TCM.Dense
    Data.TCM.Overlap

  Other-Modules:
    Data.TCM.Dense.FFI
    Data.TCM.Internal


Library utility

  Import:
    build-defaults

  HS-Source-Dirs:
    lib/utility/src

  Build-Depends:
    base                     >= 4.11      && < 5.0,
    binary                   >= 0.8       && < 1.0,
    containers               >= 0.6.2     && < 1.0,
    deepseq                  >= 1.4       && < 2.0,
    foldl                    >= 1.4       && < 2.0,
    hashable                 >= 1.3       && < 2.0,
    keys                     >= 3.12      && < 4.0,
    lens                     >= 4.18      && < 6.0,
    matrix                   >= 0.3.6     && < 0.4,
    pointed                  >= 5.0       && < 6.0,
    QuickCheck               >= 2.14      && < 3.0,
    semigroupoids            >= 5.3       && < 7.0,
    vector                   >= 0.12.0.3  && < 1.0,
    vector-binary-instances  >= 0.2       && < 1.0,
    vector-instances         >= 3.4       && < 3.5,

  If impl(ghc < 9.0)
    Build-Depends:
      integer-gmp            >= 1.0.2     && < 2.0

  Exposed-Modules:
    Data.List.Utility
    Data.Matrix.NotStupid
    Data.Vector.NonEmpty


-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
--
-- Test-suites of the sub-libraries which compose PCG/PhyG
--
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --


Test-Suite test-dynamic-character

  Import:
    build-defaults

  Main-Is:
    TestSuite.hs

  Type:
    exitcode-stdio-1.0

  HS-Source-Dirs:
    lib/dynamic-character/test

  Build-Depends:
    PHANE-alphabet,
    PHANE-dynamic-character,
    PHANE-dynamic-character-element:quickcheck,
    PhyG:dynamic-character,
    PhyG:tcm,
    base                     >= 4.11      && < 5.0,
    bimap                    >= 0.3       && < 1.0,
    containers               >= 0.6.2     && < 1.0,
    QuickCheck               >= 2.14      && < 3.0,
    tasty                    >= 1.4       && < 2.0,
    tasty-quickcheck         >= 0.10      && < 1.0,
    vector                   >= 0.12.0.3  && < 1.0,

  Other-Modules:
    DirectOptimization.Pairwise.Test
    Test.Aligners
    Test.QuickCheck.Instances.DynamicCharacter


-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --
--
-- Additional executables which exist for the one of a few select purposes:
--
--   * Generating data for PCG/PhyG input
--
--   * Debugging the component sub-libraries of PCG/PhyG and PCG/PhyG itself
--
--   * Performing correctness verification
--
--   * Stocastically searching for non-totality counterexamples
--
-- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- -- --


--Executable inspect-dynamic-character
--
--  Import:
--    build-defaults,
--
--  Scope:
--    private
--
--  Main-Is:
--    Inspect.hs
--
--  HS-Source-Dirs:
--    lib/dynamic-character/test
--
--  Build-Depends:
--    PHANE-alphabet,
--    PhyG:dynamic-character,
--    PhyG:tcm,
--    base                     >= 4.11      && < 5.0,
--    bimap                    >= 0.3       && < 1.0,
--    containers               >= 0.6.2     && < 1.0,
--    QuickCheck               >= 2.14      && < 3.0,
--    tasty                    >= 1.4       && < 2.0,
--    tasty-quickcheck         >= 0.10      && < 1.0,
--    vector                   >= 0.12.0.3  && < 1.0,
--
--  Other-Modules:
--    Test.Aligners
--    Test.QuickCheck.Instances.DynamicCharacter



Benchmark bench-memoized-hashtable

  Import:
    build-defaults

  Main-Is:
    Time.hs

  Type:
    exitcode-stdio-1.0

  GHC-Options:
    -threaded
    "-with-rtsopts=-N"

  Hs-Source-Dirs:
    lib/tcm/bench

  Build-Depends:
    PhyG:tcm,
    async,
    base                     >= 4.11      && < 5.0,
    criterion                >= 1.5       && < 2.0,
    deepseq                  >= 1.4       && < 2.0,
    mtl >= 2.3 && < 3.0,
    random,      
    vector                   >= 0.12.0.3  && < 1.0,

  Other-Modules:
    Benchmark.Internal