hamscher@HT.AI.MIT.EDU (Walter Hamscher) (12/10/87)
It seems to me that your complaint is not about Steele & the rest of the committee's unwillingness to overconstrain the language in what is still a relatively unexplored area, but rather with implementors who chose to interpret the verb `ignore' in the sense of ``the compiler or interpreter can pretend it aint there'' instead of ``the compiler doesn't have to generate special code for it''. Sort of like the difference between (declare (ignore x)) and (ignore x), if you catch my drift. In any case, since you have obviously thought some about this problem perhaps you could suggest which of the three examples you gave were the `right' ones and what the spec should have been said, keeping in mind the purpose of the definition described so succinctly in the first three pages of CLtL. Walter Hamscher