typemax(Int128) * big(1.0) => -1e+00 with 256 bits of precision

[rickhg12hs]

Is this the intended behavior?

This also is one of the causes of #5935 .

julia> versioninfo()
Julia Version 0.3.0-prerelease+1757
Commit 8854bad* (2014-02-26 04:04 UTC)
Platform Info:
  System: Linux (i686-redhat-linux)
  CPU: Genuine Intel(R) CPU           T2250  @ 1.73GHz
  WORD_SIZE: 32
  BLAS: libopenblas (DYNAMIC_ARCH NO_AFFINITY)
  LAPACK: libopenblas
  LIBM: libopenlibm

julia> typemax(Int64) * big(3.14)
-3.140000000000000124344978758017532527446746826171875e+00 with 256 bits of precision

julia> typemax(Int128) * big(3.14)
-3.140000000000000124344978758017532527446746826171875e+00 with 256 bits of precision

julia> typemax(Int64) * big(1.0)
-1e+00 with 256 bits of precision

julia> @show typemax(Int128) * big(1.0)
typemax(Int128) * big(1.0) => -1e+00 with 256 bits of precision
-1e+00 with 256 bits of precision

julia> 

[StefanKarpinski]

Nope, that's very broken. The culprit is base/mpfr.jl line 259.

[rickhg12hs]

You won't like these either then.

julia> versioninfo()
Julia Version 0.3.0-prerelease+1757
Commit 8854bad* (2014-02-26 04:04 UTC)
Platform Info:
  System: Linux (i686-redhat-linux)
  CPU: Genuine Intel(R) CPU           T2250  @ 1.73GHz
  WORD_SIZE: 32
  BLAS: libopenblas (DYNAMIC_ARCH NO_AFFINITY)
  LAPACK: libopenblas
  LIBM: libopenlibm

julia> big(typemax(Uint64)) == typemax(Uint64) * big(1.0)
false

julia> big(typemax(Uint128)) == typemax(Uint128) * big(1.0)
false

julia>