On 16 Nov (09:06:03), Nick Mathewson wrote:
On Thu, Nov 16, 2017 at 8:56 AM, David Goulet dgoulet@torproject.org wrote:
On 15 Nov (13:49:54), Nick Mathewson wrote:
[...]
On the other hand, this doesn't mean that the FIFO structure we have today is a good idea at all. It probably makes sense to use the same priority queue-based scheduler thing that we use everywhere else, but possibly with a different (inverted??) priority parameter for destroyed circuits.
(We kind of need the FIFO concept for cells afaict because of the parent relationship between cells with their digest (à la git). And that is of course per circuit.)
Are you sure? DESTROY cells aren't relay cells; they don't have relay crypto done to them, and I think it's okay to re-order them with respect to other cells. I don't think they have a digest on them, do they?
OH sorry I thought you were talking about normal circuit queue here... I mis-read.
But yes, as I mentionned in this email after, moving to a prio queue for instance has starvation implication.
Sorry! David
peace,
Nick _______________________________________________ tor-dev mailing list tor-dev@lists.torproject.org https://lists.torproject.org/cgi-bin/mailman/listinfo/tor-dev