grid-strategy
grid-strategy copied to clipboard
Decouple from matplotlib and add Bokeh for backend
This PR removes the dependency on matplotlib for creating plots. I decoupled the functions that created matplotlib figures and moved them into their own class, and had the strategy classes detect which library will be used for the backend. For each supported backend, there is a class of functions that creates a grid of figures specific to the backend library being used. For example, bokeh is the first additional backend that is being supported, and it has its own class in the backends folder that contains functions for easily adding new figures in an organized manner according to the strategy that was selected initially. This is easily extensible and now we can continue to add support for other graphing libraries in the future.
This pull request introduces 3 alerts when merging 0ea4204ac5569cc1ff35db9a8820ac6f3dc7fc96 into db12cdf788f6ddca7cec231753b2014437942db3 - view on LGTM.com
new alerts:
- 3 for Unused import
Hm.. Test failures all seem to relate to coverage. I think that this might be an issue with coverage 5.0 changing the report format.
I think it's to do with this line.
What happens if we drop it?
This pull request introduces 3 alerts when merging 69feff72f372c977839e42d8d39007dc51b600ba into db12cdf788f6ddca7cec231753b2014437942db3 - view on LGTM.com
new alerts:
- 3 for Unused import
This pull request introduces 3 alerts when merging 77e74912e6a8a157d990ddfa77a03d618e90693f into db12cdf788f6ddca7cec231753b2014437942db3 - view on LGTM.com
new alerts:
- 3 for Unused import
This pull request introduces 3 alerts when merging b5632166920dd8525c2430f80a4f31138393dcba into db12cdf788f6ddca7cec231753b2014437942db3 - view on LGTM.com
new alerts:
- 3 for Unused import
I definitely don't think you should remove the failing tests, they're super important for the review process and to make sure we don't break anything.
The housekeeping / bitrot stuff you can disable like pinning coverage or whatever, but the unit tests are critical and need to be fixed (considering that last year we had to actually write all the tests in addition to getting them working, doesn't seem like it would be too difficult to get them working again).