Using
An immutable numeric literal from a method's instructions, without source-type inference.
rawBits contains the decoded bit pattern, not the instruction's encoded bytes. For 32-bit values its upper 32 bits are zero; 64-bit values retain every bit, including the sign bit. Arithmetic lit8/lit16 operands are sign-extended to 32 bits.
Integer and floating views interpret the same bits differently. This class is not Number: its accessors do not perform arbitrary numeric conversions. No matching tolerance is applied. Exact NaN payloads are preserved by rawBits; floating operations or promotion may quiet a signaling NaN on some platforms.
Equality compares opCode and rawBits exactly, independently of the containing method or occurrence. A returned instance remains usable after the bridge closes.
Functions
Double bit view of a 64-bit const, or the decoded float view promoted to double. Arithmetic lit8/lit16 operands throw IllegalStateException.
Float bit view of a 32-bit const. Other opcodes throw IllegalStateException.
Signed 32-bit view. Throws IllegalStateException for a 64-bit literal.