[adjustable arrays?] More FORMAT

[David C. Plummer <DCP@quabbin.scrc.symbolics.com>]

    Date: Wed, 20 May 87 19:27:34 PDT
    From: edsel!bhopal!jonl@navajo.stanford.edu (Jon L White)
	...
    You can imagine how someone using a system that doesn't detect this error
    will fare when trying to run on another system that really can't adjust
    simple arrays.

"You can imagine how someone using a system that doesn't do
number-of-argument checking will fare ..."

As I recall, the number-of-argument discussion did happen several months
ago, and by the strict reading of CLtL it "is an error" to supply the
wrong number of arguments but implementations are not required to
"signal an error."  To me, this array discussion should fall in the same
class: 
 - What "is an error?"
 - What "signals an error?"
 - What implementation freedom should there be?
 - How important are some of these to various parts of the language?
   For example, I would rank number of argument checking considerably
   more important than worrying about allowing some systems to adjust
   even simple arrays.