The easiest approach is often to put Racket in its own OS-level thread, where scheme_main_setup() is given a function that receives evaluations requests and sends back results.
Currently, I don't think it's possible to call into Racket, completely leave, and call back in the way you suggest. You might imagine calling scheme_set_stack_base() on each entry to reset the start of the stack traversal on each entry, but that would create various problems with, for example, Racket continuations or threads that are captured/started in one scheme_set_stack_base() and applied/resumed in another. At Mon, 02 Dec 2013 17:12:35 +0100, Daniel Langner wrote: > Hello, > > I want to embed the racket interpreter to create a C interface for a > certain r6rs scheme library. It comprises diverse functions which return > complex scheme objects. These, in turn, may be passed to other functions > within the library (and therefore shouldn't be gc'ed along the way). > > I got it so far that I call scheme_main_setup, passing along my run > function which loads the library into the scheme environment and does > the actual scheme calls. This approach, however, isn't suitable for what > I want. When run returns, all is gone and gets garbage collected (I think). > > I'd rather have a typical interface with an initialization function > (that loads the environment and returns normally), some wrapper > functions (which call scheme_dynamic_require and scheme_apply), and a > destructor function. > > Probably I'd have to use scheme_set_stack_base rather than > scheme_main_setup. Obviously I would want to reuse the environment > throughout all calls. There hopefully is a way to prevent the garbage > collector from collecting it, and also the other scheme objects which my > library returns. > > My question is: is this feasible? How would I go about it? Thanks for > your help! > > Regards, > Daniel > ____________________ > Racket Users list: > http://lists.racket-lang.org/users ____________________ Racket Users list: http://lists.racket-lang.org/users
