[racket-dev] modules with both 'test' and 'main' submodules: looks like a Dr bug?

And, in case this wasn't already obvious to those in the know, it
appears that if you compile foo.rkt and then run "racket bar.rkt", you
get the error as well. So probably something to do with something that
happens when bytecode is unmarshalled, maybe?

(Thanks to Matthew for pointing this out.)

Robby

On Thu, Apr 26, 2012 at 8:25 PM, Robby Findler
<robby at eecs.northwestern.edu> wrote:
> If you run "raco make bar.rkt" you see the same error (and DrRacket is
> doing that for you automatically). So probably the bug lies there if
> anyone wants to investigate further.
>
> Robby
>
> On Thu, Apr 26, 2012 at 3:54 PM, John Clements
> <clements at brinckerhoff.org> wrote:
>> I'm seeing an inconsistency in the behavior of submodules that looks like a
>> bug in DrRacket (as opposed to Racket).
>>
>> First, though, it looks like running a module in DrRacket runs *both* the
>> main and test submodules, if they both exist. I don't think this behavior is
>> documented; I see this text:
>>
>> "When a module is provided as a program name to the racket executable or run
>> directly within DrRacket, if the module has a main submodule, the main
>> submodule is run after its enclosing module."
>>
>> … but no similar mention of 'test'. So I'm guessing that's just a doc bug?
>>
>> The bigger issue, though, arises when I have a file containing both 'test'
>> and 'main' submodules. It appears to me that I can't require the 'main'
>> module externally when the modules are stored in separate files, but that I
>> can when e.g. using #lang racket/load.
>>
>> So, this program:
>>
>> #lang racket/load
>>
>> (module foo racket
>> (define (g) (printf "abc\n"))
>> (define (h) (printf "def\n"))
>>
>> (module+ test
>>   (g))
>>
>> (module+ main
>>   (h))
>> )
>>
>> (module bar racket
>>   (require (submod 'foo main)))
>>
>> (require 'bar)
>>
>> … produces output "def", as I'd expect, but splitting it into two files:
>>
>> foo.rkt:
>>
>> #lang racket
>>
>> (define (g) (printf "abc\n"))
>> (define (h) (printf "def\n"))
>>
>> (module+ test
>>   (g))
>>
>> (module+ main
>>   (h))
>>
>>
>> bar.rkt:
>>
>> #lang racket
>>
>> (require (submod "foo.rkt" main))
>>
>>
>>
>> … and running "bar.rkt"  *in DrRacket* produces this message:
>>
>> require: unknown module: (submod "/private/tmp/foo.rkt" main)
>>
>>
>> Running it from the command line, though, does what I'd expect:
>>
>> pcp069222pcs:/tmp clements$ racket ./bar.rkt
>> def
>>
>>
>>
>> So it looks like DrRacket's not doing quite the right thing here.
>>
>> Apologies if this has been fixed since  5.3.0.2--2012-04-20(873c1bc/g) [3m].
>>
>> John
>>
>>
>> _________________________
>>  Racket Developers list:
>>  http://lists.racket-lang.org/dev
>>