Re: [Sbcl-devel] alien callbacks: problems statement

[Nikodemus S. <nik...@ra...>]

On 29 November 2011 15:08, Kouskoulas, Yanni A.
<Yan...@jh...> wrote:

> 1. The callback itself may have moved before invocation. The GC has no way of

Untrue. Callback trampolines are allocated in static space, and they
get updated without needing a reference to a thread.

> Is this a fair summary of the problem?

You're also overlooking the fact that even high-lever Lisp code
assumes a certain dynamic environment -- namely special variable
bindings, some of which /really/ want to be thread local.

Cheers,

 -- Nikodemus