I think special processing for lambda and fn is done too early. The reader should preserve the symbols as is.
ERR: invalid lambda expression : "lambda"
> 'fn
ERR: invalid lambda expression : "fn"
> 'a
a
> (quote lambda)
ERR: invalid lambda expression : " lambda)"
> (fn (x) x)
(lambda (x) x)
> '(fn (x) x)
(lambda (x) x)
> (legal? "fn")
true
> (legal? "lambda")
true
> (sym "fn")
fn
I think this is a bug.
symbol table should permit "fn" or "lambda" as name.
This didn't work either, surprised me:
Quote
> (define λ fn)
ERR: invalid lambda expression : " fn)"
> (define λ 'fn)
ERR: invalid lambda expression : "fn)"
> (set 'λ 'fn)
ERR: invalid lambda expression : "fn)"
> (constant 'λ 'fn)
ERR: invalid lambda expression : "fn)"
> (constant 'λ fn)
ERR: invalid lambda expression : " fn)"
The keywords
These keywords translate into a parenthesized-expression attribute. A parenthesized expression is then
> (lambda)
(lambda )
> (lambda-macro)
(lambda-macro )
>
This is also the reason that these keywords are not counted as normal symbols:
> (length (lambda))
0
>
The
Hi, Lutz.
Speed efficiency for read? or for eval? If it is for read, then I agree. But if it it not, I suggest putting preprocessing stage between read and eval stages. Then the symbols lambda, lambda-macro, fn and fn-macro can be used in a user land.
Now follow expr is ok:
> (sym "fn")
fn
> (sym "lambda")
lambda
> (sym "lambda-macro")
lambda-macro
> (lambda? (cons (sym "fn")))
nil
;; but get lambda exprssion in dynamically is not ok
> (lambda? (eval (cons (sym "fn"))))
ERR: invalid function : (fn)