Closed ba2tro closed 5 months ago
Attention: Patch coverage is 82.60870%
with 8 lines
in your changes are missing coverage. Please review.
Project coverage is 71.34%. Comparing base (
3fc5540
) to head (461ac46
).
Files | Patch % | Lines |
---|---|---|
src/ProtocolZoo/ProtocolZoo.jl | 75.75% | 8 Missing :warning: |
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
With the query specification as fifo, we don't have issues with outdated history tags and we can run Entangler and Swapper for infinite rounds. But a new issue comes out of this, where we have deadlock in the network due to a leftward node (say, node 2) having all its slots entangled with a single or multiple nodes only in the right(3, 4,...). This way, no swap can occur since we need a node in the left and one in the right to do a swap. Neither is it able to generate new entanglement with nodes on left since its slots can't be freed from entanglement with right nodes due to no swaps. If you check the log in buildkite, the end is all about entanglement attempts failing
Tests seem to be failing -- could you fix that? It makes reviewing much faster if the tests pass/
If it is because of a ton of noise in the logs due to warnings from SumTypes, you can add the same addition you had in #92
The FIFO/FILO changes look good. I have not looked into the consumer in detail yet.
This looks great! I made a couple of minor changes, hopefully did not break anything. I will plan to merge shortly.
Thanks for fixing the issues I introduced!
EntanglementConsumer protocol for network simulations. Currently it uses
query
anduntag
and we empty the qubits manually by zeroing the stateindices and setting the staterefs to nothing for the slots, just for initial testing