Issue: STREAM-CAPABILITIES (version 3)

[David A. Moon <Moon@STONY-BROOK.SCRC.Symbolics.COM>]

I hope this issue is on the agenda for the June X3J13 meeting.  KMP and
I have prepared a revised writeup which we think is ready for release.
I'd like to distribute this to X3J13 as soon as discussion, if any, in
the cleanup subcommittee is completed.

Issue:         STREAM-CAPABILITIES
References:    Standard streams (pp. 327-329)
Category:      ADDITION
Edit history:  Version 1 by Pierson 5-Jul-88, add redesign per Pitman
               Version 2 by Moon, 10-May-89, remove controversial parts
               Version 3 by Moon, 12-May-89, improve wording and examples

Problem description:

  Programs cannot currently distinguish between interactive use and batch
  (or background) use without using implementation-dependent extensions.
  For example, there is currently no way to tell whether it is useful to
  ask a question when an unexpected situation is encountered.

  Note: earlier versions of this issue tried to solve another problem
  as well.  See discussion section.

Proposal (STREAM-CAPABILITIES:INTERACTIVE-STREAM-P):

  Add the following function:

    INTERACTIVE-STREAM-P stream

    Returns T if the stream is interactive, otherwise NIL.  Signals
    an error of type TYPE-ERROR if the argument is not a stream.

    The precise meaning of INTERACTIVE-STREAM-P is implementation
    dependent, and may depend on the underlying operating system.  However
    the general intent is to distinguish between interactive and batch (or
    background or command-file) operations.

    Some examples of the things that might identify a stream as interactive
    include:
       1. The stream is connected to a person (or equivalent) in
          such a way that the program can prompt for information and
          expect to receive different input depending on the prompt.
       2. The program is expected to prompt for input and support
          "normal input editing".
       3. READ-CHAR might hang waiting for the user to type something
          instead of quickly returning a character or EOF.

    *TERMINAL-IO* might or might not be interactive.

Examples:

  (when (> measured limit)
    (let ((error (round (* (- measured limit) 100)
                        limit)))
      (unless (if (interactive-stream-p *query-io*)
                  (yes-or-no-p "The frammis is out of tolerance by ~D%.~@
                                Is it safe to proceed? " error)
                  (< error 15))  ;15% is acceptable
        (error "The frammis is out of tolerance by ~D%." error))))

Rationale:

  INTERACTIVE-STREAM-P has been proposed several times and is clearly
  needed by any program that alters its behavior depending on whether
  it is interacting with a user or running in a "batch" mode.

Current practice:

  Most implementations have this feature already, often under a
  different name.

Cost to Implementors:

  Implementations will have to support this new function.  Correct support
  will require some thought for each operating system supported.

Cost to Users:

  None, this is an upward-compatible extension.

Cost of non-adoption:

  Less featureful language. 

Performance impact:

  None.

Benefits:

  More featureful language.

Esthetics:

  More featureful language.

Discussion:

  Note that some proposed features for telling whether two streams are
  connected to the same source or sink of information have been removed
  from this version of the proposal.  These were the functions
  STREAM-SAME-SOURCE-P, STREAM-SAME-DESTINATION-P, STREAM-SOURCE-ID-LIST,
  and STREAM-DESTINATION-ID-LIST.  These could be revived in another
  proposal if desired, but Moon thought INTERACTIVE-STREAM-P was
  important and didn't want it to be lost due to controversy over these 
  unrelated functions.