GSS memory allocation initial cut for review

Nico Williams nico at
Wed Oct 5 14:25:46 EDT 2011

On Wed, Oct 5, 2011 at 12:15 PM, Sam Hartman
<hartmans at> wrote:
>    Nico> For any one mechglue implementors are free to do as they wish.
>    Nico> But I thought you wanted all mechglues on a platform to share
>    Nico> an allocator, and *that* would require some consensus among
>    Nico> implementors.
> I do, but my timely pressing concern is the changes to the MIT code.
> For now we're calling heapalloc. But if the consensus is something else
> and we've made changes to run allocations through one place it's easy to
> match that consensus.
> We presented the long-term plan only for context; the long-term
> discussion needs to happen on kitten.
> I've said this several times both on krbcore and krbdev.
> Well, OK, I may have only said it once on krbdev, but that was in the
> original message.

You know very well that I'm not on krbcore, so I've only made you
repeat yourself once :)

The original krbdev post is a bit unclear as to whether the part about
all mechglues sharing an allocator is an immediate or long-term aspect
of the proposal -- it does acknowledge the need for standardization,
so I guess that's enough to conclude the latter.

BTW, has krbcore become a function of the consortium?  With membership
a benefit to consortium members?


More information about the krbdev mailing list