Re: the global package

[BIM@MIT-AI]

Not to flame, but i am beginning to suspect that entirely too many things have been put into
the global package. Every time i turn around, some perfectly harmless function written
by gjs needs to be shadowed or renamed. Perhaps instead just the minimal lisp kernal
should be defined in global, and some other package other than si (but with a similarly
short name) should be created to hold all of the lisp machine system functions.
This comment depended on si being defined as a place where regular users
should not go. If that is not the case, then all the functions should be documented
as to their home packages. Basically, when i get a message that CONSTRUCTOR
has been illegally defined, i begin to suspect that maybe JONL wasn't completely crazy.

Second flamish issue: global special variables.
They should have a funny character on the front. This will:
1) reduce the likelihood of getting them accidently [gee, i am a random simpleminded
programmed and i think i want a special var called top-window, lets setq it to nil...]
Long lists of reserved names are too hard to remember. A * or % of & or something on the
front would win. Better yet, GET THEM OUT OF THE GLOBAL PACKAGE. They are system
variables, they should be in a system package, not the GLOBAL package which should
only contain the lisp kernal. (to repeat myself)
2) Make it easier to read code containing them.

At the risk of being thrown from the building, perhaps a multicious training _ convention would
have prevented all this trouble?

bim