Force, rentrent promise promise/e (or, pam/e is too strict)

[LeifAndersen]

I have the following program:

#lang racket

(require data/enumerate
         data/enumerate/lib)

(pam/e (λ (x) x)
       (letrec ([expr/e (delay/e
                         (or/e
                          natural/e
                          (pam/e (λ (x) x)
                                 expr/e
                                 #:contract any/c)))])
         expr/e)
       #:contract any/c)

And when I run it, I get:

. . racket/racket/collects/racket/private/promise.rkt:104:10: force: reentrant promise `promise/e'

I suspect the problem is related to the inner pam/e, because when I turn the inner expression:

(pam/e (λ (x) x)
    expr/e
    #:contract any/c)))])

Into: expr/e the program runs normally.

[LeifAndersen]

For now I found data/enumerate/unsafe, and my program seems to work fine in it.

[jeapostrophe]

A smaller program breaks too:

#lang racket/base
(require pict/tree-layout
         data/enumerate
         data/enumerate/lib)

(define bt/e
  (delay/e
   (or/e (single/e #f)
         (pam/e (λ (x) (tree-layout (car x) (cdr x)))
                (cons/e bt/e bt/e)
                #:contract tree-layout?))))

(from-nat bt/e 0)

On Wed, Mar 2, 2016 at 2:00 PM, Leif Andersen [email protected]
wrote:

For now I found data/enumerate/unsafe, and my program seems to work fine
in it.

—
Reply to this email directly or view it on GitHub
#4 (comment).

Jay McCarthy
Associate Professor
PLT @ CS @ UMass Lowell
http://jeapostrophe.github.io

       "Wherefore, be not weary in well-doing,
  for ye are laying the foundation of a great work.

And out of small things proceedeth that which is great."
- D&C 64:33

[rfindler]

I just got back to looking into this and I'm not seeing the error anymore, although I think I did see if when Leif first posted this.

Maybe one of the compiler optimization fixes fixed it?

[jeapostrophe]

I'm on commit 509da64135f412acd8522747daa3568524ac59ae and I have the error.

On Wed, Mar 9, 2016 at 6:06 PM, Robby Findler [email protected]
wrote:

I just got back to looking into this and I'm not seeing the error anymore,
although I think I did see if when Leif first posted this.

Maybe one of the compiler optimization fixes fixed it?

—
Reply to this email directly or view it on GitHub
#4 (comment).

Jay McCarthy
Associate Professor
PLT @ CS @ UMass Lowell
http://jeapostrophe.github.io

       "Wherefore, be not weary in well-doing,
  for ye are laying the foundation of a great work.

And out of small things proceedeth that which is great."
- D&C 64:33

[rfindler]

I'm two days ahead of you, but I don't really see any obvious commits that
would fix it. Nevertheless, I get the expected error for both of these
programs, namely:

delay/e-expression: broke its own contract

promised: two-way-enum?

produced: #<infinite-one-way-enum...>

in: an and/c case of

  (and/c

   infinite-enum?

   two-way-enum?

   flat-enum?)

contract from: /Users/robby/tmp.rkt

blaming: /Users/robby/tmp.rkt

(assuming the contract is correct)

at: /Users/robby/tmp.rkt:7.0

context...:

/Users/robby/git/exp/plt/racket/collects/racket/contract/private/blame.rkt:156:0:
raise-blame-error16

/Users/robby/git/exp/plt/racket/collects/racket/private/promise.rkt:104:10

/Users/robby/git/exp/plt/racket/collects/racket/private/more-scheme.rkt:265:2:
call-with-exception-handler

/Users/robby/git/exp/plt/racket/collects/racket/private/promise.rkt:96:0:
force/generic

/Users/robby/git/exp/plt/extra-pkgs/data/data-enumerate-lib/data/enumerate/private/core.rkt:1076:9

/Users/robby/tmp.rkt: [running body]

[jeapostrophe]

I get

force: reentrant promise `promise/e'
  context...:
   /Users/jay/Dev/scm/plt/racket/collects/racket/private/promise.rkt:96:0:
force/generic
   ...private/base.rkt:107:23

 /Users/jay/Dev/scm/plt/racket/collects/racket/contract/private/base.rkt:159:5

 /Users/jay/Dev/scm/plt/racket/collects/racket/contract/private/list.rkt:339:2

 /Users/jay/Dev/scm/plt/racket/collects/racket/contract/private/arrow-higher-order.rkt:454:2

 /Users/jay/Dev/scm/plt/racket/collects/racket/contract/private/arr-i.rkt:1094:2:
un-dep/maybe-chaperone
   ...rivate/arr-i.rkt:914:19
   /Users/jay/Dev/scm/plt/racket/collects/racket/private/kw.rkt:1585:36
   ...private/more.rkt:576:56

 /Users/jay/Dev/scm/plt/extra-pkgs/data-enumerate-lib/data-enumerate-lib/data/enumerate/private/more.rkt:601:13:
temp203
   /Users/jay/Dev/scm/plt/racket/collects/racket/private/promise.rkt:96:0:
force/generic

 /Users/jay/Dev/scm/plt/extra-pkgs/data-enumerate-lib/data-enumerate-lib/data/enumerate/private/core.rkt:1074:9
   /tmp/t.rkt: [running body]

 /Users/jay/Dev/scm/plt/extra-pkgs/compiler-lib/compiler-lib/compiler/commands/test.rkt:176:16
t.rkt: raco test: test raised an exception

Compilation exited abnormally with code 1 at Wed Mar  9 19:04:46

If you expect these programs to both error, how do you think we should
write down recursive one-way enumerations?

Jay

On Wed, Mar 9, 2016 at 7:01 PM, Robby Findler [email protected]
wrote:

I'm two days ahead of you, but I don't really see any obvious commits that
would fix it. Nevertheless, I get the expected error for both of these
programs, namely:

delay/e-expression: broke its own contract

promised: two-way-enum?

produced: #<infinite-one-way-enum...>

in: an and/c case of

(and/c

infinite-enum?

two-way-enum?

flat-enum?)

contract from: /Users/robby/tmp.rkt

blaming: /Users/robby/tmp.rkt

(assuming the contract is correct)

at: /Users/robby/tmp.rkt:7.0

context...:

/Users/robby/git/exp/plt/racket/collects/racket/contract/private/blame.rkt:156:0:
raise-blame-error16

/Users/robby/git/exp/plt/racket/collects/racket/private/promise.rkt:104:10

/Users/robby/git/exp/plt/racket/collects/racket/private/more-scheme.rkt:265:2:
call-with-exception-handler

/Users/robby/git/exp/plt/racket/collects/racket/private/promise.rkt:96:0:
force/generic

/Users/robby/git/exp/plt/extra-pkgs/data/data-enumerate-lib/data/enumerate/private/core.rkt:1076:9

/Users/robby/tmp.rkt: [running body]

—
Reply to this email directly or view it on GitHub
#4 (comment).

Jay McCarthy
Associate Professor
PLT @ CS @ UMass Lowell
http://jeapostrophe.github.io

       "Wherefore, be not weary in well-doing,
  for ye are laying the foundation of a great work.

And out of small things proceedeth that which is great."
- D&C 64:33

[rfindler]

I still don't see why the error went away for me, but the obvious thing to guess is that something happened in the two days of git commits. And yes, I did indeed see that error before.

Meanwhile, here's the program I think that you wanted?

#lang racket/base
(require pict/tree-layout
data/enumerate
data/enumerate/lib)

(define bt/e
(delay/e
 (or/e (single/e #f)
       (pam/e (λ (x) (tree-layout (car x) (cdr x)))
              (cons/e bt/e bt/e)
              #:contract tree-layout?))
 #:two-way-enum? #f))

(for/list ([i (in-range 1 10)])
  (binary-tidier (from-nat bt/e i)))