[stack] syntax preference question

Everyone likes debating matters of syntax, right?

Here's a (rather monolithic) way of writing the quadratic formula in  
Factor:

   [ rot 4 * * [ [ neg ] [ sq ] bi ] dip - sqrt [ + ] [ - ] 2bi ] keep
   2 * tuck [ / ] 2bi@

Here's the same formula in Fifth using c-expressions. This is the  
syntax I've been using for my examples recently:

   keep(rot 4 * * dip(bi(neg, sq)) - sqrt 2bi(+, -)) 2 * tuck 2bi@(/)

Here's the same thing using s-expressions. Note that because  
whitespace can't be used as composition in all cases, we need an  
explicit composition combining form (called 'o'):

   (o (keep (o rot 4 * * (dip (bi neg sq)) - sqrt (2bi + -)) 2 * tuck  
(2bi@ /))

This could be improved somewhat by making [a b c] special syntax for  
(o a b c):

   [(keep [rot 4 * * (dip (bi neg sq)) - sqrt (2bi + -)] 2 * tuck  
(2bi@ /)]

Anyway, I'm unsure which of the two approaches, c-expressions or s- 
expressions, is preferable. C-expressions look a fair bit cleaner (as  
concatenation always represents composition), but they're less regular  
(code is no longer just a simple tree). There may also be some  
advantages to making composition explicit of which I'm not yet aware.

Yet another thought was to have a more "special" syntax with square  
brackets used for the cleave combinator (with a number prefixed for  
keeping more than one value per quotation) and things like `foo(a) as  
syntactic sugar for dip(foo(a)):

  [rot 4 * * `[neg, sq] - sqrt 2[+, -], 2 * tuck 2@[/]]

Of course, this approach could be made more severe.

Any thoughts?

- John