Issue PRINTER-WHITESPACE Writeup

Issue:        PRINTER-WHITESPACE

Forum:	      X3J13 Letter Ballot

References:   22.1.3.8 (pp22-7..22-8),

	      22.1.3.10 (p22-9)

Category:     CLARIFICATION

Edit history: 16-Jun-93, Version 1 by Pitman

Status:       Proposal JUST-ONE-SPACE passed (7+1)-3 on letter ballot 93-304



Problem Description:


  In the descriptions of how lists and vectors print, the draft 

  standard says "whitespace" is printed where CLtL1 says "a space".

  Is this an unintentional change?


Proposal (PRINTER-WHITESPACE:JUST-ONE-SPACE):


  When *print-pretty* is false, require the printer to use exactly

  one space in the indicated situations.  In general, when *print-pretty*

  is false, recommend that implementations and users output a minimum

  of space in this situation.  [This is status quo under CLtL1]


  Rationale: This forces implementations to use very little space

             and makes printer behavior more regular across implementations.


Proposal (PRINTER-WHITESPACE:WHITESPACE):


  Affirm that where the printer is defined to output whitespace,

  arbitrary whitespace is permissible.  Recommend that implementations

  and users use a minum of whitespace when *print-pretty* is false,

  but permit flexibility.  [This is status quo under dpANS]


  Rationale: This makes it less likely that users will write code

	     that is not robust in the face of varying amounts of 

	     whitespace inserted by entities other than the printer

	     (e.g., a human being).


Test Cases:


  #1: (CHAR (FORMAT NIL "~S" '(A B)) 2)

  might return #\Space in some implementations and #\Newline in others.


  #2: (CHAR (FORMAT NIL "~S" '(A B)) 3)

  might return #\Space in some implementations and #\B in others.


  #3: (LENGTH (FORMAT NIL "~S" '(A B)))

  might return 5 in some implementations or greater numbers in others.


Current Practice:


  Not provided.


Cost to Implementors:


  JUST-ONE-SPACE: Relatively small.

  WHITESPACE:     None. (status quo)


Cost to Users:


  JUST-ONE-SPACE: None.

  WHITESPACE:     None. (status quo)


Cost of Non-Adoption:


  See Rationale above.


Benefits:


  See above.


Editorial Impact:


  Small.


Aesthetics:


  Some people might quibble a little, but mostly pretty neutral.


Discussion:


  This is in response to Public Review comment Loosemore #15.


  Pitman prefers option WHITESPACE.


  Pitman believes the complained-about change occurred during 

  the integration of the pretty printer description with the 

  printer chapter.  At Waters' direction,  he allowed this change 

  to be taken as editorial--probably a bad idea in retrospect.