Origin of eval()-ing in separate namespace object

[Jussi Piitulainen]

rusi writes:

> On Sunday, December 8, 2013 4:05:54 PM UTC+5:30, Kalinni Gorzkis wrote:
>
> > By which languages(s) Python was inspired to support evaluating
> > expressions and executing statements in a separate "namespace"
> > object?
> 
> > This syntax:
> > eval(expression,globals) or exec(code,globals)
> > What is the origin of the functionality provided by the globals
> > argument?
> 
> Been here since the days of scheme at least
> http://docs.racket-lang.org/guide/eval.html#%28part._namespaces%29
> 
> For the record lisp was conceptualized in the late 50s and
> implemented by 1960.  By the 80s it was widely regarded as the
> premier AI language but it was also clear to users that the scoping
> rules were terribly wrong.  So a number of the then lisps coalesced
> and re-separated into 2 major dialects -- scheme and common lisp.
> 
> I expect it -- 2 argument eval -- goes all the way back to the
> earliest lisp but Ive not access to the history.

Yes. From p. 13 of LISP 1.5 Programmer's Manual (the preface is dated
in 1962):

# _evalquote_ is defined by using two main functions, called _eval_
# and _apply_. _apply_ handles a function and its arguments, while
# _eval_ handles forms. Each of these functions also has another
# argument that is used as an association list for storing the values
# of bound variables and function names.

That association list is the namespace. _eval_ and _apply_ pass it
around and extend it as they call each other to evalute code.

So it seems that at least the idea of an explicit namespace argument
was there from the start. Mistakes related to the original dynamic
scoping continued to be made.