On Sun, Dec 6, 2020, at 15:44, John Benediktsson wrote:
> We need to figure that out.
>
> bi/bi*/bi@
> tri/tri*/tri@
> cleave/spread/apply
>
> It might make more sense to have a numerical syntax instead of words or
> better yet would be to only have cleave/spread/apply and have shorter
> symbolic syntax for it.
>
> Plenty of places where [ foo ] keep bar seems to look cleaner than [ foo ] [
> bar ] bi and I’d rather we had more consistency and fewer tokens possibly.
I'm speaking purely for myself, but even to the extent I might agree in the
narrow case of [ foo ] [ bar ] bi versus [ foo ] keep bar, the relationship
between bi/bi*/bi@ and tri/tri*/tri@ I think is important, and can cause me to
lean for bi in many cases even where a keep sequence might arguably be clearer.
That said, if cleave/spread/apply didn't feel so "heavy", I'd definitely be
fine ditching the bi/tri families in favor of the generics. Your point on
syntax might work. It also might honestly not feel heavy in practice, even
today; it'd be interesting to try going a couple weeks using *only*
cleave/spread/apply directly and see if it actually makes my code worse.
_______________________________________________
Factor-talk mailing list
Factor-talk@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/factor-talk
