Re: Parenthesized Target Notation

On Mon, 2005-05-02 at 14:03 +1200, Richard A. O'Keefe wrote:

> ECMA/ETL3 is absolutely clear that the newline has no special significance
> that a blank does not have.  Omitting semicolons only when they would have
> been at the end of a line is a style rule, nothing more.

In 1998 we persuaded Bertrand Meyer to use newline for disambiguation in
this case. Here's what Bertrand wrote on 1998-08-10 in comp.lang.eiffel:

   On paper it is not too difficult to add a rule that 
   two successive instructions or assertions must be separated
   by a newline if there is no semicolon. This may mess up the
   compilers (in ISE's implementation we very early in the
   game transform newlines into generic whitespace) but that's
   not a really good excuse as long as the change is implementable
   in the end, which it is. I'll raise the issue with the NICE
   language committee. 

   On the one hand, I don't think it's a matter of life and
   death ... On the other hand, if we can remove a source of
   potential error, however unrealistic, we should do it. 

However, the issue was never presented to the NICE language committee.

Regards,
Roger
-- 
Roger Browne <[email protected]>