That has a couple potential problems, including breaking tail calls, but is also potentially faster. This is all discussed in the paper I linked.
Sam On Sat, Aug 10, 2019, 12:16 AM Hendrik Boom <hend...@topoi.pooq.com> wrote: > On Fri, Aug 09, 2019 at 01:03:17PM -0400, Sam Tobin-Hochstadt wrote: > > This is possible -- the central trick is to maintain information about > > the call stack separately, perhaps by using continuation marks. > > Or set a global variable on entry, and reset it on exit. > If on entry the global variable is set (before you set it, of course), > you've called yourself, directly of indirectly. > > > > > We (mostly Phil Nguyen) recently built a tool that does this. If you > > take your program, install the "termination" package, and then add: > > > > (require termination) > > (begin/termination (recur1 0)) > > > > you get this error message: > > > > possible-non-termination: Recursive call to `#<procedure:recur1>` has > > no obvious descent on any argument > > - Preceding call: > > * 1st arg: 3 > > - Subsequent call: > > * 1st arg: 6 > > > > We have a paper, here: https://arxiv.org/abs/1808.02101 which goes > > into detail about how it works. The source code is here: > > https://github.com/philnguyen/termination > > > > Sam > > > > On Fri, Aug 9, 2019 at 11:58 AM Kees-Jochem Wehrmeijer > > <henc...@gmail.com> wrote: > > > > > > Hi, > > > > > > I'm wondering whether it's possible for a function to detect whether > it's being called by itself, possibly indirectly. I want to use this to > prevent endless loops and generate different results in these situations. > For example, imagine I have the following functions: > > > > > > (define (foo loc) > > > (list > > > (cons loc 'a) > > > (cons (+ loc 1) 'b) > > > (cons (+ loc 2) 'c))) > > > > > > (define (combine a b) > > > (lambda (loc) > > > (append (a loc) (b (+ loc 3))))) > > > > > > So I can do: > > > > ((combine foo foo) 0) > > > => '((0 . a) (1 . b) (2 . c) (3 . a) (4 . b) (5 . c)) > > > > > > Now, let's say I have these recursive definitions: > > > > > > (define (recur1 loc) > > > ((combine foo recur2) loc)) > > > > > > (define (recur2 loc) > > > (recur1 loc)) > > > > > > Defined in this way, a call to (recur1 0) loops endlessly. I would > like it too return something like > > > > (recur1 0) > > > => '(0 . a) (1 . b) (2 . c) (loop 0)) > > > > > > So I was thinking if I could make it work if I could do something like: > > > (define (recur1 loc) > > > (if (in-recursive-call?) > > > (list (cons 'loop 0)) > > > ((combine foo recur2) loc))) > > > > > > Is that something that's possible? Is there a better way to do what I > want? > > > > > > Thanks, > > > Kees > > > > > > -- > > > You received this message because you are subscribed to the Google > Groups "Racket Users" group. > > > To unsubscribe from this group and stop receiving emails from it, send > an email to racket-users+unsubscr...@googlegroups.com. > > > To view this discussion on the web visit > https://groups.google.com/d/msgid/racket-users/292977cf-7800-453f-9d4b-b6a1ee525287%40googlegroups.com > . > > > > -- > > You received this message because you are subscribed to the Google > Groups "Racket Users" group. > > To unsubscribe from this group and stop receiving emails from it, send > an email to racket-users+unsubscr...@googlegroups.com. > > To view this discussion on the web visit > https://groups.google.com/d/msgid/racket-users/CAK%3DHD%2Bb0odK1qfox9Ah2umLqKxUhas_GTJUv95Avm536g9ybGg%40mail.gmail.com > . > > -- > You received this message because you are subscribed to the Google Groups > "Racket Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to racket-users+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/racket-users/20190809182512.cffth5bwreff5urs%40topoi.pooq.com > . > -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to racket-users+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/CAK%3DHD%2BZCpv-x9Cj00vMkj__pNmCH6eKrOE-XVf0GZH0dDxoNow%40mail.gmail.com.
