Deprecate VirtualGraphComposite
#532
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #532 +/- ##
==========================================
- Coverage 89.65% 86.71% -2.94%
==========================================
Files 24 24
Lines 1760 1762 +2
==========================================
- Hits 1578 1528 -50
- Misses 182 234 +52 ☔ View full report in Codecov by Sentry. |
| "of newer QPUs and in future will raise an exception; if needed, " | ||
| "follow the instructions in the shimming tutorial at " | ||
| "https://github.com/dwavesystems/shimming-tutorial instead. ", | ||
| DeprecationWarning |
There was a problem hiding this comment.
I think we might need to set the stacklevel for the warning to be shown from the correct code line?
There was a problem hiding this comment.
I saw that we are inconsistent. LeapHybridDQMSampler sets the stacklevel but LazyEmbeddingComposite does not. I think we need that to override Python's newer default that doesn't show warnings unless users set them to show. Do we want to do that?
| @@ -32,14 +37,19 @@ | |||
| from dwave.system.composites.embedding import FixedEmbeddingComposite | |||
| from dwave.system.flux_bias_offsets import get_flux_biases | |||
|
|
|||
| from warnings import warn | |||
There was a problem hiding this comment.
Python libraries get imported first.
Also, IMO it's nicer to just import warnings then do warnings.warn for a slightly cleaner namespace.
| Deprecated. | ||
| Virtual graphs are deprecated due to improved calibration of newer QPUs; to | ||
| calibrate chains for residual biases, follow the instructions in the | ||
| `shimming tutorial <https://github.com/dwavesystems/shimming-tutorial>`_. | ||
|
|
There was a problem hiding this comment.
Consider using .. deprecated:: version Sphinx directive.
There was a problem hiding this comment.
I did that now for the class (line 50 below).
|
We should update the tests with assertWarns for the deprecation warning. I know we already have a few warnings when we run the tests, but I'd like to not make it worse. |
Closes #531 and #54