28-bit two’s complement integer. The sign bit is stored redundantly in the top 5 bits of the word.
The sign bit is stored as part of the type code, allowing a 28 bit signed short float format. The format of short floating point numbers is:
0 3 4 5 12 13 31 --------------------------------------------------------------- | Type code (4) | Sign (1) | Exponent (8) | Mantissa (19) | ---------------------------------------------------------------
The floating point number is the same format as the IBM RT PC supports for single precision numbers, except it has been shifted right by four bits for the type code. The result of any operation is therefore truncated. Long floating point numbers are also available if you need more accuracy and better error propagation properties.
character object holding a character code, control bits, and font in the following format:
0 4 6 7 8 15 16 23 24 31 --------------------------------------------------------------- | Type code (5) | Unused (3) | Font (8) | Bits (8) | Code (8) | ---------------------------------------------------------------
Used to mark the presence of multiple values on the stack. The low 16 bits indicate how many values are being returned. Note that only 65535 values can be returned from a multiple-values producing form. These are pushed onto the stack in order, and the Values-Marker is returned in register A0.
Object used as the catch tag for unwind-protects. Special things happen when a catch frame with this as its tag is encountered during a throw. See Catch for details.
Illegal object trap. This value is used in symbols to signify an undefined value or definition.
Object used to mark a miscop as interruptible. This object is put in one of the registers and signals to the interrupt handler that the miscop can be interrupted safely. Only miscops that can take a long time (e.g., length when passed a circular list, system call miscops that may wait indefinitely) are marked this way.