Email list hosting service & mailing list manager


Re: when GC is permitted Tom Lord 15 Jan 2004 18:04 UTC

    > From: Eric Knauel <xxxxxx@informatik.uni-tuebingen.de>

    > I'm very confident that migrating those bindings to the SRFI-FFI is
    > not much work.  Checking whether the GC annotations are (still)
    > correct and a few search/replace-operations should be enough.

At least for the regex1.c and syscalls1.c code, that's false.

Not that I think checking the GC annotations is a small task -- I
think the fact that you already say that's necessary raises the
question of the cost (in labor and impact on the world) of converting
to the draft vs the cost of converting to a more general FFI.

But the nested cons bug in syscalls1.c, the nested enter_fixnum bug in
regex1.c, and the boolean comparisons using C operators in syscalls1.c
illustrate that beyond search/replace and tweaking GC annotation
expressions need to be examined individually and in some cases
rewritten, including a need for new (gc protected) local variable
declarations.

-t