change arg type to atom in print and trace

[noskill]

with this change one can print what actually is during expression evaluation

this program will print now (woman $Y) vs printing False without these changes

(= (woman $X) False)
(= (sisters (woman $X) (woman $Y)) (trace! (woman $Y) True))
!(let $W (sisters $X $Y) $W)

[Necr0x0Der]

I considered this before, but decided not to change the argument type of println! to Atom, because it is not too usable:

• if you want to print the result of some calculations, they will be reduced in any case.
• if you want to avoid wrapping println! into let each time you want to print the results of calculations, the type of println! argument should not be Atom. And this is quite a common use cae.
  For example, if the argument type is Atom, and we want to print get-atoms result, ! (println! (get-atoms &self)) will print (get-atoms &self), which is useless, while ! (let $atoms (get-atoms &self) (println! $atoms)) will still output the reduced atoms. Thus, Atom-type argument makes only sense to print the exact provided expression. This is a less frequent case than printing the result of evaluation. If one wants to print an unreduced expression, it can be quoted, e.g.

(: TRACE (-> Atom Atom))
(= (woman $X) False)
(= (sisters (woman $X) (woman $Y)) (trace! (TRACE (woman $Y)) True))
!(let $W (sisters $X $Y) $W)

I'm not sure, maybe printing atoms as is in case of trace! can be more demanded. I don't have a strong opinion here. But it looks questionable for println!. @vsbogd , @luketpeterson , do you have any opinion?

[vsbogd]

I agree with @Necr0x0Der arguments about println! it is probably not convenient to use let each time and usually you need to reduce atom before printing.

Regarding trace! I would ask @ngeiswei opinion because he introduced trace! and probably most active user of it. I almost always use trace! to print atom without reducing it, and I use quoting to reach it. If it is more frequent use-case then I think making first argument of trace! atom may help.

[noskill]

Quotation solves problem for me, i tried quote from standard library before doing PR but it didn't work

[Necr0x0Der]

But I'd also like to hear @ngeiswei opinion. Maybe, Atom in trace! can indeed be convenient.

[ngeiswei]

It looks like I'm always using trace! with fully reduced first argument. And one can always wrap a let around trace! to reduce its first argument if needed. So I guess it wouldn't hurt to have its type be Atom. But Just keep in mind that it is not how Idris or Haskell trace works.

Maybe we should consider introducing quotes instead of having to decide at the type level how a function should behave.

[Necr0x0Der]

OK, if @ngeiswei has no preferences for Atom type, then let's keep it as is. Some official quoting operation in stdlib would be useful for sure.