Function not giving expected output Eos smart contract

Question

My contract .Cpp part

void mainloan::adduwr(name acc_name, double interest, uint64_t loan_amount )
{
  print("Adding underwriter", acc_name);
  // require_auth( _self );

  uwr_table.emplace(get_self(), [&](auto &u) {
    u.acc_name = acc_name;
    u.interest_rate =interest;
    // u.loan_amount=loan_amount;
    u.giving_loan_amunt = loan_amount;
  });
}

.hpp part

 struct [[eosio::table]] underwriter_info
  {
    name acc_name;
    // std::vector<uint64_t> interest_rate;
    // std::vector<uint64_t> loan_amount; 
      double interest_rate;
      uint64_t giving_loan_amunt;
    auto primary_key() const
    {
      return acc_name.value;
    }
  };

  typedef eosio::multi_index<"underwriter"_n, underwriter_info> underwriter;

  underwriter uwr_table;


  [[eosio::action]] 
  void adduwr(name acc_name,double interest, uint64_t loan_amount);

Cleos call

guru@ubuntu:~/contracts/mainloan$ cleos --url http://jungle2.cryptolions.io:80 push action microfinanc1 adduwr ['lender',8.6,10000] -p microfinanc1
executed transaction: 26b8218bbcd623430b440efecfcbb395cc5430f25d16900497e96c006c78977f  120 bytes  210 us
#  microfinanc1 <= microfinanc1::adduwr         {"acc_name":"lender","interest":"8.59999999999999964","loan_amount":10000}
>> Adding underwriterlender
warning: transaction executed locally, but may not be confirmed by the network yet         ] 

so the problem happens when i called the action

cleos --url http://jungle2.cryptolions.io:80 push action microfinanc1 adduwr ['lender',8.6,10000] -p microfinanc1

output from cleos

{"acc_name":"lender","interest":"8.59999999999999964","loan_amount":10000}

so 8.6 is chnaged to 8.59999999999999964 why this is happening? if anyone have any suggestion to give,please say it.

Answer 1

The builtin floating-point types store values as finite binary fractions. For technical specification, see Wikipedia. For a human-readable explanation of the why, a good starting point is The Floating-Point Guide.

The number 8.6 is not representable as a finite binary fraction. So, the double type stores the closest possible approximation instead, which is displayed to you as 8.59999999999999964.