[Gc] Re: [bdwgc] Export mark-bit manipulation functions.
ivmai at mail.ru
Sat Jan 14 11:42:25 PST 2012
13 01 2012, 22:54 Petter Urkedal <reply+c-865805-42298043e682fa7591f1579b64dcbf2b6c086712-460469 at reply.github.com>:
> Hi Ivan,
> On 2012-01-13, Ivan Maidanski wrote:
> > The latter seems to me not logically correct as the object is marked (somehow - it is implementation-dependent).
> My take on it is that GC_set_mark_bit and GC_clear_mark_bit does not
> change the object as seen from the users. Rather they change the
> collectors idea about its reachability, which is dependent on an
> external link to the object, rather than the content of the object. The
> way I use it myself, objects are never changed once constructed, except
> possibly for an internal next-link.
Hmm... Looks reasonable, thank you for the explanation. I agree with you now.
> > I've committed this patch (adding const to GC_is_marked).
> Thanks, I don't mind the others being non-const. I only call this from
> low-level allocation and clean-up code, which already contain uintmax_t
> to void * casts.
> Reply to this email directly or view it on GitHub:
More information about the Gc