Replace use of fdlibm with openlibm.

Makefile

@@ -44,7 +44,7 @@ install: release
  cp $(BUILD)/bin/*julia* $(PREFIX)/bin
  cd $(PREFIX)/bin && ln -s julia-release-$(DEFAULT_REPL) julia
  cp -R -L $(BUILD)/lib/julia/* $(PREFIX)/lib/julia
- -cp $(BUILD)/lib/lib{Rmath,amd,amos,arpack,cholmod,colamd,fdm,fftw3,fftw3f,fftw3_threads,fftw3f_threads,glpk,glpk_wrapper,gmp,gmp_wrapper,grisu,history,julia-release,$(OPENBLASNAME),openlibm,pcre,random,readline,suitesparse_wrapper,tk_wrapper,umfpack,z}.$(SHLIB_EXT) $(PREFIX)/lib
+ -cp $(BUILD)/lib/lib{Rmath,amd,amos,arpack,cholmod,colamd,openlibm,fftw3,fftw3f,fftw3_threads,fftw3f_threads,glpk,glpk_wrapper,gmp,gmp_wrapper,grisu,history,julia-release,$(OPENBLASNAME),openlibm,pcre,random,readline,suitesparse_wrapper,tk_wrapper,umfpack,z}.$(SHLIB_EXT) $(PREFIX)/lib
 # Web-REPL stuff
  -cp $(BUILD)/lib/mod* $(PREFIX)/lib
  -cp $(BUILD)/sbin/* $(PREFIX)/sbin

base/float.jl

@@ -70,7 +70,7 @@ iround(::Type{Uint64}, x::Float64) = box(Uint64,fpuiround64(unbox(Float64,x)))
 # TODO: Int128
 
 # this is needed very early because it is used by Range and colon
-floor(x::Float64) = ccall(dlsym(_jl_libfdm,:floor), Float64, (Float64,), x)
+floor(x::Float64) = ccall(dlsym(libopenlibm,:floor), Float64, (Float64,), x)
 
 iceil(x::FloatingPoint) = itrunc(ceil(x)) # TODO: fast primitive for iceil
 ifloor(x::FloatingPoint) = itrunc(floor(x)) # TOOD: fast primitive for ifloor

base/openlibm.jl

@@ -0,0 +1,157 @@
+libopenlibm = dlopen("libopenlibm")
+
+macro libopenlibmfunc_1arg_float(T,f)
+ quote
+ $(esc(f))(x::Float64) = ccall(dlsym(libopenlibm,$(string(f))), Float64, (Float64,), x)
+ $(esc(f))(x::Float32) = ccall(dlsym(libopenlibm,$(string(f,"f"))), Float32, (Float32,), x)
+ $(esc(f))(x::Real) = ($f)(float(x))
+ @vectorize_1arg $T $f
+ end
+end
+
+macro libopenlibmfunc_1arg_float(T,f)
+ quote
+ $(esc(f))(x::Float64) = ccall(dlsym(libopenlibm,$(string(f))), Float64, (Float64,), x)
+ $(esc(f))(x::Float32) = ccall(dlsym(libopenlibm,$(string(f,"f"))), Float32, (Float32,), x)
+ $(esc(f))(x::Real) = ($f)(float(x))
+ @vectorize_1arg $T $f
+ end
+end
+
+macro libopenlibmfunc_1arg_int(T,f,name...)
+ if length(name)>0
+ fname = name[1]
+ else
+ fname = f
+ end
+ quote
+ $(esc(fname))(x::Float64) = ccall(dlsym(libopenlibm,$(string(f))), Int32, (Float64,), x)
+ $(esc(fname))(x::Float32) = ccall(dlsym(libopenlibm,$(string(f,"f"))), Int32, (Float32,), x)
+ @vectorize_1arg $T $fname
+ end
+end
+
+macro libopenlibmfunc_2arg(T,f)
+ quote
+ $(esc(f))(x::Float64, y::Float64) = ccall(dlsym(libopenlibm,$(string(f))), Float64, (Float64, Float64,), x, y)
+ $(esc(f))(x::Float32, y::Float32) = ccall(dlsym(libopenlibm,$(string(f,"f"))), Float32, (Float32, Float32), x, y)
+ @vectorize_2arg $T $f
+ end
+end
+
+@libopenlibmfunc_1arg_float Number cbrt
+@libopenlibmfunc_1arg_float Number sin
+@libopenlibmfunc_1arg_float Number cos
+@libopenlibmfunc_1arg_float Number tan
+@libopenlibmfunc_1arg_float Number sinh
+@libopenlibmfunc_1arg_float Number cosh
+@libopenlibmfunc_1arg_float Number tanh
+@libopenlibmfunc_1arg_float Number asin
+@libopenlibmfunc_1arg_float Number acos
+@libopenlibmfunc_1arg_float Number atan
+@libopenlibmfunc_1arg_float Number asinh
+@libopenlibmfunc_1arg_float Number acosh
+@libopenlibmfunc_1arg_float Number atanh
+@libopenlibmfunc_1arg_float Number log
+@libopenlibmfunc_1arg_float Number log2
+@libopenlibmfunc_1arg_float Number log10
+@libopenlibmfunc_1arg_float Real log1p
+@libopenlibmfunc_1arg_float Real logb
+@libopenlibmfunc_1arg_float Number exp
+@libopenlibmfunc_1arg_float Real expm1
+@libopenlibmfunc_1arg_float Number erf
+@libopenlibmfunc_1arg_float Number erfc
+@libopenlibmfunc_1arg_float Real ceil
+@libopenlibmfunc_1arg_float Real floor
+#@libopenlibmfunc_1arg_float Real rint
+@libopenlibmfunc_1arg_float Number lgamma
+
+@libopenlibmfunc_1arg_float Number sqrt
+@libopenlibmfunc_1arg_float Number exp2
+#@libopenlibmfunc_1arg_float Real nearbyint
+@libopenlibmfunc_1arg_float Real trunc
+@libopenlibmfunc_1arg_float Real round
+
+@libopenlibmfunc_2arg Number atan2
+atan2(x::Real, y::Real) = atan2(float64(x), float64(y))
+@libopenlibmfunc_2arg Number hypot
+hypot(x::Float32, y::Float64) = hypot(float64(x), y)
+hypot(x::Float64, y::Float32) = hypot(x, float64(y))
+
+gamma(x::Float64) = ccall(dlsym(libopenlibm, :tgamma), Float64, (Float64,), x)
+gamma(x::Float32) = float32(gamma(float64(x)))
+gamma(x::Real) = gamma(float(x))
+@vectorize_1arg Number gamma
+
+lfact(x::Real) = (x<=1 ? zero(x) : lgamma(x+one(x)))
+@vectorize_1arg Number lfact
+
+max(x::Float64, y::Float64) = ccall(dlsym(libopenlibm, :fmax), Float64, (Float64,Float64), x, y)
+max(x::Float32, y::Float32) = ccall(dlsym(libopenlibm, :fmaxf), Float32, (Float32,Float32), x, y)
+@vectorize_2arg Real max
+
+min(x::Float64, y::Float64) = ccall(dlsym(libopenlibm, :fmin), Float64, (Float64,Float64), x, y)
+min(x::Float32, y::Float32) = ccall(dlsym(libopenlibm, :fminf), Float32, (Float32,Float32), x, y)
+@vectorize_2arg Real min
+
+#@libopenlibmfunc_1arg_int Real lrint
+#@libopenlibmfunc_1arg_int Real lround iround
+function ilogb(x::Float64)
+ if x==0 || isnan(x)
+ throw(DomainError())
+ end
+ int(ccall(dlsym(libopenlibm,:ilogb), Int32, (Float64,), x))
+end
+function ilogb(x::Float32)
+ if x==0 || isnan(x)
+ throw(DomainError())
+ end
+ int(ccall(dlsym(libopenlibm,:ilogbf), Int32, (Float32,), x))
+end
+@vectorize_1arg Real ilogb
+
+@libopenlibmfunc_1arg_float Real significand
+
+ldexp(x::Float64,e::Int) = ccall(dlsym(libopenlibm, :ldexp), Float64, (Float64,Int32), x, int32(e))
+ldexp(x::Float32,e::Int) = ccall(dlsym(libopenlibm, :ldexpf), Float32, (Float32,Int32), x, int32(e))
+# TODO: vectorize does not do the right thing for these argument types
+#@vectorize_2arg Real ldexp
+
+begin
+ local exp::Array{Int32,1} = zeros(Int32,1)
+ global frexp
+ function frexp(x::Float64)
+ s = ccall(dlsym(libopenlibm,:frexp), Float64, (Float64, Ptr{Int32}), x, exp)
+ (s, int(exp[1]))
+ end
+ function frexp(x::Float32)
+ s = ccall(dlsym(libopenlibm,:frexpf), Float32, (Float32, Ptr{Int32}), x, exp)
+ (s, int(exp[1]))
+ end
+ function frexp(A::Array{Float64})
+ f = similar(A)
+ e = Array(Int, size(A))
+ for i = 1:numel(A)
+ f[i] = ccall(dlsym(libopenlibm,:frexp), Float64, (Float64, Ptr{Int32}), A[i], exp)
+ e[i] = exp[1]
+ end
+ return (f, e)
+ end
+ function frexp(A::Array{Float32})
+ f = similar(A)
+ e = Array(Int, size(A))
+ for i = 1:numel(A)
+ f[i] = ccall(dlsym(libopenlibm,:frexpf), Float32, (Float32, Ptr{Int32}), A[i], exp)
+ e[i] = exp[1]
+ end
+ return (f, e)
+ end
+end
+
+modf(x) = rem(x,one(x)), trunc(x)
+
+^(x::Float64, y::Float64) = ccall(dlsym(libopenlibm, :pow), Float64, (Float64,Float64), x, y)
+^(x::Float32, y::Float32) = ccall(dlsym(libopenlibm, :powf), Float32, (Float32,Float32), x, y)
+
+^(x::Float64, y::Integer) = x^float64(y)
+^(x::Float32, y::Integer) = x^float32(y)

base/start_image.jl

@@ -8,9 +8,8 @@ _jl_lib = ccall(:jl_load_dynamic_library,Ptr{Void},(Ptr{None},),C_NULL)
 # Essential libraries
 libpcre = dlopen("libpcre")
 libgrisu = dlopen("libgrisu")
-_jl_libm = dlopen("libm")
-_jl_libfdm = dlopen("libfdm")
 librandom = dlopen("librandom")
+libopenlibm = dlopen("libopenlibm")
 @windows_only _jl_advapi32 = dlopen("Advapi32")
 
 # Optional libraries

base/sysimg.jl

@@ -54,7 +54,7 @@ include("operators.jl")
 include("pointer.jl")
 
 _jl_lib = ccall(:jl_load_dynamic_library,Ptr{Void},(Ptr{None},),C_NULL)
-_jl_libfdm = dlopen("libfdm")
+libopenlibm = dlopen("libopenlibm")
 
 include("float.jl")
 include("reduce.jl")
@@ -105,7 +105,7 @@ include("client.jl")
 include("intfuncs.jl")
 include("floatfuncs.jl")
 include("math.jl")
-include("math_libm.jl")
+include("openlibm.jl")
 include("sort.jl")
 include("combinatorics.jl")
 include("statistics.jl")