Add method to compile geb to boolean circuits

[mariari]

This is just #105 rebased

[AHartNtkn]

Looks like the code I wrote.

[mariari]

Might be good to have a followup pr changing this to an actual CL bitvector. they have many efficient operations and can be easily done

https://novaspec.org/cl/f_bit-and

for an example of some oeprations on this. Esp if we want to add more operations

https://novaspec.org/cl/t_bit-vector

[AHartNtkn]

Looks fine

[AHartNtkn]

Most of this is formatting and import stuff that I'm not qualified to comment on. The only substantial difference from my original code is the to-bitc refactoring. After a bit of squinting and local testing, I can see it does the same thing as the original code. It certainly looks nicer and after a bit of discussion with @mariari, we got it in a state I can approve of

[mariari]

I noticed a bug in one of the extractions

GEB-TEST> 
(bitc:to-circuit (geb:to-bitc (uncurry geb-bool:bool
                                       geb-bool:bool
                                       geb-bool:or))
                 :or)
def or x1 x2 = {
  ((1 - x1) * 1) + (x1 * x2)
};
GEB-TEST> (dom (uncurry geb-bool:bool geb-bool:bool geb-bool:or))
(× GEB-BOOL:BOOL GEB-BOOL:BOOL)
GEB-TEST>
(gapply (uncurry geb-bool:bool geb-bool:bool geb-bool:or)
        (list (left so1)
              (right so1)))
(right s-1)
GEB-TEST>
(gapply (uncurry geb-bool:bool geb-bool:bool geb-bool:or)
        (list (right so1)
              (left so1)))
(right s-1)
GEB-TEST> 
(gapply (uncurry geb-bool:bool geb-bool:bool geb-bool:or) (list (left so1)
                                                                (left so1)))
(left s-1)
GEB-TEST> (geb.main::reducer (uncurry geb-bool:bool geb-bool:bool geb-bool:or))
(∘ (case (∘ (<-left GEB-BOOL:BOOL GEB-BOOL:BOOL)
            (<-left (× GEB-BOOL:BOOL GEB-BOOL:BOOL) GEB-BOOL:FALSE))
         (∘ (<-right GEB-BOOL:BOOL GEB-BOOL:BOOL)
            (<-left (× GEB-BOOL:BOOL GEB-BOOL:BOOL) GEB-BOOL:TRUE)))
   (dist (× GEB-BOOL:BOOL GEB-BOOL:BOOL) GEB-BOOL:FALSE GEB-BOOL:TRUE)
   ((∘ (GEB-BOOL:BOOL (∘ GEB-BOOL:TRUE (->1 GEB-BOOL:BOOL)))
       (<-left GEB-BOOL:BOOL GEB-BOOL:BOOL))
    (<-right GEB-BOOL:BOOL GEB-BOOL:BOOL)))

https://pomf2.lain.la/f/1pm7rire.png

Here I have an overcomplicated or function seen by the image above.

it seems that it evaluates fine with the interpreter I have, however it seems to produce the wrong vampir circuit where if both booleans are 0 it produces . I assume something simple is wrong

[rokopt]

This is great -- I'd favor having it merged as soon as the conflicts are resolved. Thanks!

[mariari]

This is great -- I'd favor having it merged as soon as the conflicts are resolved. Thanks!

conflicts get resolved on merge, the conflicts aren't real, the PR is final as is, and will be merged as is, unless we find a bug or changes are requested

[agureev]

Not currently of importance, but somewhere down the line probably a good idea to include some more documentation specifying the passes this makes.

[mariari]

Would be good, and a documentation string on each instance as well to explain how it can be used.

I may want to open a second PR that is based on this and the lambda changes that removes the trans ordering

[rokopt]

it seems that it evaluates fine with the interpreter I have, however it seems to produce the wrong vampir circuit where if both booleans are 0 it produces . I assume something simple is wrong

Was this one fixed in one of the later commits?

[mariari]

it seems that it evaluates fine with the interpreter I have, however it seems to produce the wrong vampir circuit where if both booleans are 0 it produces . I assume something simple is wrong

Was this one fixed in one of the later commits?

This was fixed in the first commit by AHartNtkn after it was brought to his attention