documentation mistake

[Syntacticlosure]

In the doc of function get-default-percentages in gui-lib:

method(send a-panel:dragable get-default-percentages subwindow-count)
--
→ (listof (and/c real? (between/c 0 1)))
subwindow-count : exact-positive-integer?

Should the contract of function result be (listof (and/c real? exact? (between/c 0 1))) , not (listof (and/c real? (between/c 0 1))) ?

[bennn]

It looks ok to me. I can see it might be a bad idea to have inexact percentages, but it doesn't seem so bad that it should be prohibited.

[Syntacticlosure]

@bennn The problem is , if u override this function with augment and return an inexact real number between 0 and 1 , u will get an error.

[bennn]

I don't understand. Can you post the code you have?

In the implementation, get-default-percentages checks that the percentages add to 1. That might error if some percentages are inexact, but it might not:

Welcome to Racket v7.0.0.2.
> (= 1 (apply + '(1/5 0.2 0.6)))
#t
> (= 1 (apply + (list 1/3 (exact->inexact 2/3))))
#t

[Syntacticlosure]

My code is here:

(require racket/gui framework)
(define f (new frame% [label "test"][width 400][height 200]))

(define p (new (class
                   panel:horizontal-dragable%
                 (super-new)
                 (define/augment (get-default-percentages i)
                   (match i
                     [2 (list #e0.5 #e0.5)]
                     [_ (build-list i (λ (_) (/ 1 i)))])))
               [parent f]))

(define m1 (new message% [label "left"][parent p]))
(define m2 (new message% [label "right"][parent p]))
(send f show #t)

It runs perfectly but if you replace #e0.5 with 0.5 ,you will get this:

container-redraw: result from place-children is not a list of length 2 (matching the input list length) whose elements are lists of length 4 of exact integers
result: '((0 0 190.0 162) (195.0 0 190.0 162))

[LeifAndersen]

Given that they must add up to exactly 1, it makes sense to me that they should be limited to exact-real? numbers.

But it does seem like there is also a bug in the implementation (or contract) this function uses as well.