0.5.1

New Features

• Add support for dwave-cloud-client~=0.13.0.

0.5.0.post0

Relax upper bound for the dwave-cloud-client dependency.

0.5.0

Summary

Add support for fast anneal protocol and improve performance.

New Features

• Add support for fast anneal protocol. Display the new solver parameter fast_anneal and the new solver property fast_anneal_time_range in the visualizer. See #170

• Replace builtin JSON encoder with orjson, achieving 2-5x speed-up of problem and solver API endpoints.

• Replace WSGIAsyncServer with dwave.cloud.auth.server.BackgroundAppServer, a more robust and stable multi-threaded server. By using a multi-threaded server we improve the visualizer app load time, as browsers can download static files and make API requests in parallel now. See #165.

Upgrade Notes

• dwave.inspector.utils.NumpyJSONProvider is removed and no longer available. Use orjson.dumps with option=orjson.OPT_SERIALIZE_NUMPY instead.

• Minimal version of dwave-cloud-client required is now 0.11.0 to support #165 and fix #164.

Bug Fixes

• Omit legacy solver parameters, like num_spin_reversal_transforms and postprocessing params (postprocess, beta, chains) from the visualizer. See #166 and #168.

• Improve performance of the inspector server by postprocessing solver data inplace, saving one deep copy of solver metadata (~600kB for Advantage) on each API request of solver data.

0.4.4

New Features

• Add Python 3.12 support.

0.4.3

New Features

• Handle web browser requests to the inspector server with problem ID missing by redirecting to the last problem show() was called for.

  Besides being practical, this mitigates issues in some environments/setups with URL rewriters/proxies dropping URL path and/or query.

Upgrade Notes

• Upgrade your python to 3.8+. We no longer support python 3.7 and below.

0.4.2

Bug Fixes

• Use vcrpy-compatible version of urllib3 (<2) for testing in #157
• Migrate to codecov's orb from the deprecated uploader in #158
• Update examples to use newer QPU in #155

0.4.1

Upgrade Notes

• The .show() call does not return a RichDisplayURL anymore. Instead, we reverted to returning URL in a plain str. Jupyter viewer will display the inspector inline if possible, otherwise the URL is opened in a browser.

  To render the inspector URL inline in a Jupyter notebook cell, independently of the .show() call, you can do:

  url = dwave.inspector.show(...)
  ...
  from dwave.inspector.utils import RichDisplayURL
  RichDisplayURL(url)
  

Bug Fixes

• Fix duplicated inline render of the Inspector in GUI Jupyter in case when the returned (rich URL) object is displayed in the same cell the dwave.inspector.show() was called. See #152

0.4.0

New Features

• Add support for jupyter-server-proxy. Inspector URL gets rewritten prior to opening/viewing according to external URL as defined by the new config environment variable: DWAVE_INSPECTOR_JUPYTER_SERVER_PROXY_EXTERNAL_URL. See #144

• Add support for Python 3.11

• Add support for dimod~=0.11.0 and dimod~=0.12.0

• Add support for running the inspector behind a network proxy. Now users can register a URL rewriter specific to their environment using the inspectorapp_proxies entry point. See #141

• Use relative API paths to enable app hosting on externally-configured paths (app#110). See #146.

• Render inspector inline in Jupyter notebooks. See #109 and #133.

Upgrade Notes

• Drop Python 3.6 support

• Drop dimod~=0.8.0 and dimod~=0.9.0 support

• Custom viewer can now return False to signal a non-blocking show() behavior is desired. Previously the value returned was ignored by the caller.

• Lower bound on Flask version is now 2.2.

Bug Fixes

• Improve error handling and display in dwave-inspectorapp (app#100)

• Make show() non-blocking regardless of the block argument when no viewer manages to open the inspector page. See #139.

• Upgraded JSON serialization to use Flask's new DefaultJSONProvider (introduced in Flask 2.2). The "old" way is deprecated by Flask in 2.2, to be dropped in next minor release, 2.3.

0.3.0

New Features

• Add support for Zephyr topology. See #137.

0.2.11

New Features

• Add support for (serialization of) NumPy types used anywhere in the inspector data served from the Inspector web server (Flask app). See #81.

Bug Fixes

• Use of embeddings with NumPy types was not previously possible. Now it is. See #127.