lens-transform should be consistent with lens-set about lens HOT 6 CLOSED

Or, does the argument order of lens-set* and lens-transform* make more sense? (target first, then lens, then new-value/transformer-function)

It breaks the convention of having the lens as the first argument, but allows the lens and the new-value/transformer-function to be next to each other, and is more consistent with hash-set, hash-update, and other similar functions.

from lens.

I'm not sure how to fold plurality into argument conventions, because there's several plural cases:

• Multiple lenses, one target, one view (lens-view*)
• Multiple lenses, one target, multiple views (lens-transform*, lens-set*)
• One lens, multiple targets, one view (possibly something like lens-set-all, lens-transform-all)
• One lens, multiple targets, multiple views (possibly something like lens-set-each, lens-transform-each)
• Multiple lenses, multiple targets, one view (possibly something like lens-set-all for multiple lenses)

And probably more I haven't thought of. The singular lens-view couldn't be both consistent with a function that takes multiple lenses and a function that takes multiple views if variable-arity functions are desired and the rest arguments are uniform.

from lens.

Well, right now, the argument orders for the single-lens cases of lens-view*, lens-set* and lens-transform* are consistent with each other and consistent with functions like hash-ref, hash-set, and hash-update.

If you thought it would make more sense, you could change all of them to follow that convention, but that breaks the convention of having the lens as the first argument, and also turns it into a postfix-ish feel.

I'm not sure what you mean by lens-view being consistent with a function that takes multiple views though?

from lens.

I meant lens-set, my apologies.

from lens.

Maybe worth noting: For the plural of lens-set, I debated between two possibilities:

• (lens-set* target lens ... new-val), equivalent to (lens-set (lens-thrush lens ...) target new-val)
• (lens-set* target lens new-val ... ...), equivalent to (for/fold ([target target]) ([lens/val (in-slice 2 ....)]) (lens-set lens target val)), filling in the gaps.

I chose the second one because I thought it would be more useful, especially since it's easy to throw a lens-thrush in there for the first one, and also because the second would be consistent with hash-set*.

from lens.

Let's move general discussion of argument order to #73

from lens.