Destroying the stack and creating a new on on address change sounds like
the right thing to do to me, so long as the effect of doing so is contained
to the stack -- ie., higher layer and application state in the unikernel
isn't / doesn't have to be affected.

If someone wants fancier behaviour I guess you'd look to extract the state
from the stack as required, and then recreate the new stack from that --
cf. the Irmin/ARP prototype, or the Irmin/TCP prototype...

On 13 July 2016 at 16:30, Mindy Preston notifications@github.com wrote:

That sounds like pretty resounding "yes, I think?" to "is it fine?" (Right
now we don't do even close to the right thing with DHCP, just getting a
lease once and assuming it's valid forever, so the question is rather
hypothetical I admit.)

—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
#557 (comment), or mute
the thread
https://github.com/notifications/unsubscribe/AAKP4IlNdgGfU8kbw6h1gKvuKZB2V-9Hks5qVQSzgaJpZM4JLf8G
.

Richard Mortier
mort@cantab.net

arp and ip module types should assume one ip address and default gateway #557

Description

Metadata

Metadata

Assignees

Labels

Type

Projects

Milestone

Relationships

Development

Issue actions