[racket-dev] [racket] Code compiled for version ~a, not ~a

From: Guillaume Marceau (gmarceau at gmail.com)
Date: Sat Sep 17 23:36:57 EDT 2011

On Sat, Sep 17, 2011 at 11:11 PM, Robby Findler
<robby at eecs.northwestern.edu> wrote:
> Whoops, I see we lost the mailing list at some point in this thread.

We're back now.


> If the former, then those are the directories that DrRacket
> intentionally doens't compile (and those are the ones that I'd expect
> you to have multiple copies of).

That's pretty subtle. I don't remember seeing a notice of this in the
documentation. Is it somewhere and I didn't stumble on it? It might be
worth changing the behavior or the error message to make the behavior
more discoverable.


> Does this mean that you added it to the directory that's listed first
> in the PLTCOLLECTS variable or that you added a second path to that
> variable?

It breaks both ways, with

  export PLTCOLLECTS=';c:\documents\collects'

and with

   export PLTCOLLECTS='c:\documents\collects;'


> If the latter, what does
>
>  #lang racket/base
>  (require setup/dirs)
>  (find-collects-dir)
>
> produce on your machine? (And what is PLTCOLLECTS)?

It produced #<path:C:\documents\projects\plt\collects> , which is
where my git repository is.

In case it is useful, (current-library-collection-paths) produces:

'(#<path:C:\Documents and Settings\gmarceau\Application
Data\Racket\5.1.3.9\collects>
  #<path:C:\documents\projects\plt\collects>
  #<path:c:\documents\collects>)




>> When logging-tool is linked with raco, the error doesn't occur at
>> lunch because DrRacket doesn't attempt loads tools the linked
>> collects.
>
> It is supposed to. So maybe something is wrong there too.

ah, ok.



Posted on the dev mailing list.