Closed randomir closed 1 year ago
Merging #478 (d02e37d) into master (44de5dc) will decrease coverage by
4.15%
. The diff coverage is100.00%
.
@@ Coverage Diff @@
## master #478 +/- ##
==========================================
- Coverage 90.29% 86.14% -4.16%
==========================================
Files 24 24
Lines 1669 1674 +5
==========================================
- Hits 1507 1442 -65
- Misses 162 232 +70
Impacted Files | Coverage Δ | |
---|---|---|
dwave/system/samplers/dwave_sampler.py | 83.72% <100.00%> (-1.31%) |
:arrow_down: |
dwave/system/coupling_groups.py | 12.50% <0.00%> (-81.25%) |
:arrow_down: |
dwave/system/samplers/leap_hybrid_sampler.py | 61.42% <0.00%> (-14.29%) |
:arrow_down: |
dwave/system/samplers/clique.py | 73.85% <0.00%> (-13.73%) |
:arrow_down: |
dwave/system/composites/embedding.py | 96.57% <0.00%> (-1.15%) |
:arrow_down: |
dwave/system/testing.py | 95.23% <0.00%> (-0.80%) |
:arrow_down: |
:mega: We’re building smart automated test selection to slash your CI/CD build times. Learn more
I'll merge this for now as is, and try to generalize it a bit in follow-up PR.
Fixes #477, with a caveat -- user-supplied (or default num qubits-based) solver preference is ignored on
trigger_failover()
. We overrideorder_by
in order to de-prioritize failed solvers.In summary, this is a quickfix for #477, and a nicer fix should respect the lower-level solver preference, but penalize failed solvers. This seems hard to implement in dwave-system without a lot of code duplication from dwave-cloud-client. One possible, and pretty general, approach is to apply
order_by
transformations sequentially (in the cloud-client), one per config level, instead of higher-level setting overridding a lower-level setting.