[no subject]

[Richard M. Stallman <RMS at MIT-AI>]

A long time ago I implemented PACKAGE-DECLARE to operate
at macro expansion time.  Since this is the only function
I ever implemented to work that way, I must have had some
reason in mind, but I can't remember what it is.

Does anyone have any idea what good this could do?

Since people seem to be complaining about it,
if nobody can think of a reason to keep it this way,
I propose to change it to be an ordinary FEXPR.

Or perhaps it should expand into (EVAL-WHEN (COMPILE EVAL LOAD)
around a call to a fexpr.  Then it would be possible to have
a PACKAGE-DECLARE actually followed by code in that package
(necessarily put into that package by package prefixes),
and things will work properly on readfiling, fasloading or
compiling that file in a world in which the package is not yet defined.

The comments in front of the definition of PACKAGE-DECLARE
are no help; they seem to be false.