gives incorrect results (Int64 overflow) or errors (for BigInt)

[lampretl]

The following code

using IntegerSmithNormalForm
X = BigInt[  3   2  -4   0  -2   6  -5  -3   6   9;
            -7   5  -7   7  -9   8   9   4   1   5;
             5  -3  -6  -1   0   1   4   3  -4   9;
            -6   9   8   2   8   8  -8   4  -3   5;
            -8   8   9   4   6  -4   5   1  -1   5;
             1  -2   4  -6   6  -8   1   2   5   2;
            -5  -6  -5   2  -9   7  -7  -6   3   2;
             2   0  -9   0   8   1   7  -1  -9  -3;
            -2  -1   9  -3  -8  -4  -8   4  -5   0;
            -5  -5  -1   9  -4  -8  -1  -9   1   0 ];   
P,D,Q = snf(Int64.(X));   println(D==P*Int64.(X)*Q, " ", big.(D)==big.(P)*X*big.(Q))

shows that the results are correct in Int64 arithmetic with overflows, but incorrect in the exact BigInt arithmetic. Running the command with exact integers gives an error:

P,D,Q = snf(X);   println(D==P*X*Q)
ERROR: MethodError: no method matching addRow(::LinearAlgebra.Adjoint{BigInt, Matrix{BigInt}}, ::Int64, ::Int64, ::Int64)
The function `addRow` exists, but no method is defined for this combination of argument types.

Closest candidates are:
  addRow(::AbstractMatrix{E}, ::Any, ::Any, ::E) where E
   @ IntegerSmithNormalForm ~/.julia/packages/IntegerSmithNormalForm/6WZhT/src/IntegerSmithNormalForm.jl:67

Stacktrace:
 [1] snf!(A::Matrix{BigInt})
   @ IntegerSmithNormalForm ~/.julia/packages/IntegerSmithNormalForm/6WZhT/src/IntegerSmithNormalForm.jl:197
 [2] snf(A::Matrix{BigInt})
   @ IntegerSmithNormalForm ~/.julia/packages/IntegerSmithNormalForm/6WZhT/src/IntegerSmithNormalForm.jl:238
 [3] top-level scope
   @ REPL[5]:1

In SNF, the matrix entries typically become enormous during the computation. Without support for BigInt, the results are useless.