At the weekly call yesterday, I said that it would be somewhat complicated, but upon further thought, I saw that it is, in fact, easy. The earliest deadline first view will enable the user to avoid lower-priority timed-out frames from building up beyond higher-priority non-timed-out frames while the queue cannot make progress. The cost is 16 bytes of memory per queue item.
At the weekly call yesterday, I said that it would be somewhat complicated, but upon further thought, I saw that it is, in fact, easy. The earliest deadline first view will enable the user to avoid lower-priority timed-out frames from building up beyond higher-priority non-timed-out frames while the queue cannot make progress. The cost is 16 bytes of memory per queue item.