[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]


I've thought about FUNCTION-TYPE on and off over the summer, as one of
the more important issues for us to revisit.  I've changed my mind; I am
now in favor of STRICT-REDEFINITION as long as the proposal that
specifies that it "is an error" to give symbols to functions that expect
functions. ) as long as the restriction is "is an error" rather than
"signals an error".

I think that the proposal has technical (linking, analysis) and
aesthetic (cleaner language semantic) advantages, and that backward
compatibility is well-served by noting that many implementations will
continue to support automatic coercion.

I think we owe X3J13 a new draft of the proposal with fewer (rather than
more) alternatives expressed. 

All of the alternatives expressed so far have some flaws. I'm willing to
work on the STRICT-REDEFINITION proposal to correct its problems (e.g.,
deal with *MACROEXPAND-HOOK*, add COERCE to type FUNCTION).

Are there any volunteers to work on any of the other proposals?