Literals

There  are  three  types  of  literal  constant  -  numerals,  character
constants, and string constants.

Numerals are written in the following style
        12  5237563  24.6  4.5e13  0.63e-6
A numeral containing either a decimal point or a scale  factor  (`.'  or
`e')  is  fractional,  and  is  stored  internally  as  double precision
floating point (accuracy approximately 17 decimal digits).  Integers are
held   in  a  different  internal  representation,  and  have  unbounded
precision.

The two kinds of number, integer and fractional, are both of type `num',
as  far  as  the  type-checker  is concerned, and can be freely mixed in
calculations.  There is automatic conversion from integer to  fractional
when  required,  but  not  in  the  opposite direction.  To convert from
fractional to integer, use `entier' (see standard environment).

Negative numbers are denoted by applying the prefix `-'  operator  to  a
numeral, thus:
        -12  -4.5e13
but  note  that  the notation -12 is an expression, not a literal, so if
you wish to apply a function to it, you must write f (-12), not  f  -12,
which would be read as an attempt to subtract 12 from f.

Character constants are written using single quotes, thus
        'a'    '0'    '\n'

String constants are written using double quotes, thus
        "hello dolly"    ""    "\n\n\n"

Escape  conventions  in  character  and  string constants are as in `C',
using the backslash character.
        \n    newline
        \t    tab
        \f    formfeed
        \r    carriage return
        \b    backspace
        \\    the backslash character itself
        \'    single quote
        \"    double quote
        \ddd  up to three decimal digits, arbitrary ascii character code

Note that literal newlines are not allowed inside  character  or  string
quotes,  unless  escaped  by  backslash,  in  which  case the newline is
ignored.  Thus
        "hello \
dolly" means the same as "hello dolly".

These conventions are exactly as in `C',  except  that  we  use  decimal
rather  than  octal  for  numeric  specification  of  an arbitrary ascii
character code (the use of  octal  numbers  in  computing  now  seems  a
curious anachronism which there is no need to perpetuate).