rspec/rules/S2164/rule.adoc

== Why is this an issue?

For small numbers, ``++float++`` math has enough precision to yield the expected value, but for larger numbers, it does not. ``++BigDecimal++`` is the best alternative, but if a primitive is required, use a ``++double++``.


=== Noncompliant code example

[source,text]
----
float a = 16777216.0f;
float b = 1.0f;
float c = a + b; // Noncompliant; yields 1.6777216E7 not 1.6777217E7

double d = a + b; // Noncompliant; addition is still between 2 floats
----


=== Compliant solution

[source,text]
----
float a = 16777216.0f;
float b = 1.0f;
BigDecimal c = BigDecimal.valueOf(a).add(BigDecimal.valueOf(b));

double d = (double)a + (double)b;
----


=== Exceptions

This rule doesn't raise an issue when the mathematical expression is only used to build a string.

[source,text]
----
System.out.println("["+getName()+"] " +
           "\n\tMax time to retrieve connection:"+(max/1000f/1000f)+" ms.");
----


== Resources

* https://wiki.sei.cmu.edu/confluence/x/CtcxBQ[CERT, FLP02-C.] - Avoid using floating-point numbers when precise computation is needed