Matthew, hello. On 2010 Jun 12, at 13:53, Matthew Flatt wrote:
> The problem may be that you are compiling the extension for CGC, but > you're linking and running for 3m. As a result, some local pointers in > the implementation of the extension are invisible to the GC. Aha, it would have taken a while for this to occur to me. Thanks for that. The problem doesn't magically go away, however: > % "/Data/LocalApplications/Racket/Racket v5.0/bin/raco" exe --cgc -o > build/simbad-lod server.ss > create-embedding-executable: can't find Racket executable for variant cgc > > === context === > /Data/LocalApplications/Racket/Racket > v5.0/collects/compiler/embed-unit.rkt:1008:4: core > /Data/LocalApplications/Racket/Racket > v5.0/collects/compiler/commands/exe.rkt: [running body] > /Data/LocalApplications/Racket/Racket v5.0/collects/raco/raco.rkt: [running > body] > /Data/LocalApplications/Racket/Racket v5.0/collects/raco/main.rkt: [running > body] So something may yet be amiss. But don't worry about it on my behalf: this is a sign. It's a sign that it's probably about time that I converted this extension to use the 3m GC, which will evade the problem in the Correct fashion. > Along those lines, my guess is that requiring `web-server/servlet' > changes the timing of a GC so that one occurs during the initialization > of the extension. Pointers then get lost in a way that drops the > instantiation of the extension-implemented module. Hmm: the GC still retains something of the occult in my perception. Best wishes, Norman -- Norman Gray : http://nxg.me.uk _________________________________________________ For list-related administrative tasks: http://lists.racket-lang.org/listinfo/users
