re: Emery's Lexical Q

[Mendal@SIERRA.STANFORD.EDU (Geoff Mendal)]

We batted Dave Emery's lexical question (... + 1:= ...) around
Stanford and came up with two interpretations:

  (1) The declaration is syntactically illegal because it requires a
      separator to distinguish between the three lexical tokens.
  (2) The declaration is syntactically legal because even though
      the colon is the replacement character for the pound sign,
      "1:=" is not itself a legal token and therefore requires
      different interpretation.

The relevant section in the LRM appears to be 2.2(2).  The question
that Emery's example poses is:

  Must a compiler try all possible reinterpretations if it finds
  a syntax error?  Or is there something inherent in the syntax
  of Ada which requires a separator in this case?

The Stanford boys are deadlocked, and different compilers give different
answers (Verdix rejects, DEC-Ada accepts), so we'd appreciate hearing
from anyone who knows the "correct" answer.

gom
-------