I'm not sure, but I think I can be content with passing Racket bytes across the foreign interface boundary for the C runtime to use given the u8vector equivalence. Only thing I'm not clear on: Is it a good idea for me to designate the byte string as a _gcpointer on the "*" marked line, or does it not matter?
(define-ffi-definer define-ff (ffi-lib doesnt-matter)) (define-ff fill_up (_fun _pointer _int -> _void)) ; * (define size 2) (define buffer (make-bytes size)) (fill_up buffer size) On 5/23/21 8:45 PM, Philip McGrath wrote: > On Sun, May 23, 2021 at 7:57 PM Matthew Flatt <[email protected]> wrote: > >> At Sun, 23 May 2021 14:57:42 +0000, Sage Gerard wrote: >>> Is there a cross-VM way to pass an arbitrary input port to the C >>> runtime (e.g. via open-input-bytes), such that the C runtime can read >>> bytes on its own? >> >> No, not unless you know that the port's implementation is sufficiently >> constrained. In general, reading from a port can involve thread >> switches and synchronization, and those are not allowed in callbacks >> from foreign libraries. > > Off hand, I think some some version of "sanitizing" an arbitrary port for use > from C is used by: > > - `racket/system` > ([here](https://github.com/racket/racket/blob/448b77a6629c68659e1360fbe9f9e1ecea078f9c/racket/collects/racket/system.rkt#L98-L114) > and > [here](https://github.com/racket/racket/blob/448b77a6629c68659e1360fbe9f9e1ecea078f9c/racket/collects/racket/system.rkt#L173-L193)), > where the implementation comes from `racket/private/streams`; > - `racket/draw`, for the various parts of Cairo that can read from or write > to ports, in `racket/draw/unsafe/callback` (and maybe something similar in > `racket/draw/private/write-bytes`?); and > - `readline`, which uses callbacks rather than creating file-stream ports. > > Would it make sense to expose some part of this functionality as a new > primitive? > > (I'm not familiar with all of the constraints in these use-cases: maybe > they're different enough that the answer is "no".) > > -Philip > -- > You received this message because you are subscribed to the Google Groups > "Racket Users" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to [email protected]. > To view this discussion on the web visit > [https://groups.google.com/d/msgid/racket-users/010001799bd5ab46-d39f83a5-bb2d-49e5-8ce2-3315307742ac-000000%40email.amazonses.com](https://groups.google.com/d/msgid/racket-users/010001799bd5ab46-d39f83a5-bb2d-49e5-8ce2-3315307742ac-000000%40email.amazonses.com?utm_medium=email&utm_source=footer). -- ~slg -- You received this message because you are subscribed to the Google Groups "Racket Users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/racket-users/31b6053a-e1ed-3cd8-2e2b-d9ebb5a086d5%40sagegerard.com.
