Re: Issue: EVAL-WHEN-NON-TOP-LEVEL (Version 5)

[sandra%defun@cs.utah.edu (Sandra J Loosemore)]

Well, I was hoping the person who has made the most noise in the past
about the nesting behavior would speak up again in his own defense. :-)

Here's my own position on the issue, which you can quote me on if you
want. 

  In the processing situations that are controlled by EVAL and LOAD, the
  absence of the situation from an outer EVAL-WHEN means that that
  processing simply will not be performed on its body, regardless of
  whether the body contains nested EVAL-WHENs which do specify the
  appropriate situation.  However, if an outer EVAL-WHEN does not
  specify COMPILE, this does not necessarily suppress all compile-time
  evaluation of its body, since EVAL-WHENs nested inside the body which
  do specify the COMPILE situation can also cause compile-time
  evaluation.  I think this asymmetry in the handling of the COMPILE
  situation is confusing and unaesthetic. 
  
  I am also concerned about the complexity of the specification for the
  behavior of EVAL-WHEN.  I believe it is possible to come up with a
  much simpler specification that also "fixes" the nesting problem. 

As I said in my original reply, I'm willing to work on such an
alternate proposal if others are interested in seeing it, but if
everybody else is already perfectly happy with your new proposal I
won't try to obstruct it.

-Sandra
-------