Path to dynamic-place misbehaves when in executable

[pmatos]

Initially I discussed with @mflatt the issue of passing a modpath created with define-runtime-module-path-index to dynamic-place. @mflatt suggested using collapse-module-path-index before passing it to dynamic-place, which work unless you are in compiled code.

For example:
main.rkt:

#lang racket

(require racket/runtime-path
         syntax/modcollapse)

(define-runtime-module-path-index mod-for-place '(file "place-print.rkt"))

(void 
  (dynamic-place
   (collapse-module-path-index mod-for-place)
   'print))

(printf "Place created.~n")

place-print.rkt:

#lang racket

(provide print)

(define (print ch)
  (printf "Hello from another place~n"))

It works in the command line:

$ racket main.rkt
Place created.

But when it comes to running in a compiled environment:

$  raco exe -o pl-print.exe main.rkt
$  ./pl-print.exe 
dynamic-place: not a filesystem or predefined module-path
  module path: ''#%embedded:g3861:place-print
  context...:
   /home/pmatos/installs/racket-7.0/collects/racket/place.rkt:120:0: start-place*
   /home/pmatos/installs/racket-7.0/collects/racket/place.rkt:93:0: dynamic-place7
   '#%mzc:main: [running body]
   for-loop
   run-module-instance!125
   perform-require!78
   top-level: [running body]
   loop
   [repeats 1 more time]

If this is a really tricky one, is there a workaround I can use in the meantime?

[pmatos]

@mflatt I have a feeling you have been in a path related to this issue before resulting in commit 3bffcae given the commit title.

[mflatt]

Thanks! It looks like I had done most of the work to wire things up in v7.0, but I forgot to get rid of a stub is_predefined_module_path and call scheme_is_predefined_module_path. That latter function is almost (but not quite) ready to work.

[pmatos]

Thanks for the fix. It works.