Distinctions Among the "write" Predicates

• With write and writeq the term is written with respect to current operator declarations (See ref-syn-ops, for a discussion of operators).

  write_canonical(Term) writes Term to the current or specified output stream in standard syntax (see ref-syn on Prolog syntax), and quotes atoms and functors to make them acceptable as input to read/[1,2]. That is, operator declarations are not used and compound terms are therefore always written in the form:

            predicate_name(arg1, ..., argn)
            

• Atoms output by write/[1,2] cannot in general be read back using read/[1,2]. For example,

            | ?- write('a b').
            a b
            

  If you want to be sure that the atom can be read back by read/[1,2], you should use writeq/[1,2], or write_canonical/[1,2], which put quotes around atoms when necessary, or use write_term/[2,3] with the quoted option set to yes.

• write/[1,2] and writeq/[1,2] treat terms of the form '$VAR'(N) specially: they write A if N=0, B if N=1, ...Z if N=25, A1 if N=26, etc. Terms of this form are generated by numbervars/3 (see ref-lte-anv). Terms of the form '$VAR'(X), where X is not a number are written as unquoted terms. For example,

            | ?- writeq(a('$VAR'(0),'$VAR'('Test'))).
            a(A,Test)
            

  write_canonical/1 does not treat terms of the form '$VAR'(N) specially. It writes square bracket lists using ./2 and [] (that is, [a,b] is written as .(a,.(b,[]))). If the character_escapes flag is on then write_canonical/1 tries to write layout characters (except ASCII 9 and ASCII 32) in the form \lower-case-letter, if possible; otherwise, write_canonical/1 writes the \^control-char form. If the character_escapes flag is off then it writes the actual character, without using an escape sequence (see ref-syn-ces).

• Depending upon whether character escaping is on or off, writeq/[1,2] and write_canonical/[1,2] behave differently when writing quoted atoms. If character escaping is on:
  1. The characters with ASCII codes 9 (horizontal tab), 32 (space), and 33 through 126 (non-layout characters) are written as themselves.
  2. The characters with ASCII codes 8, 10, 11, 12, 13, 27, and 127 are written in their \lowercase letter form.
  3. The character with ASCII code 39 (single quote) is written as two consecutive single quotes.
  4. The character with ASCII code 92 (back slash) is written as two consecutive back slashes.
  5. All other characters are written in their \^control char form.

  If character escaping is off:

  1. The character with ASCII code 39 (single quote) is written as two consecutive single quotes.
  2. All other characters are written as themselves.
  3. In general, one can only read (using read/[1,2]) a term written by write_canonical/[1,2] if the value of the character_escapes flag is the same when the term is read as when it was written.