Skip to content

Update minimum bokeh version message#7172

Merged
jrbourbeau merged 1 commit intodask:mainfrom
jrbourbeau:fix-conda-install-bokeh
Oct 28, 2022
Merged

Update minimum bokeh version message#7172
jrbourbeau merged 1 commit intodask:mainfrom
jrbourbeau:fix-conda-install-bokeh

Conversation

@jrbourbeau
Copy link
Copy Markdown
Member

Currently we're giving conflicting version for installing with conda (1.0) and pip (2.1.1). This PR makes it so we specify the minimum bokeh version once and then use that value where needed.

@pavithraes
Copy link
Copy Markdown
Member

@jrbourbeau Shall we add the maximum bokeh version in this same PR?

@github-actions
Copy link
Copy Markdown
Contributor

Unit Test Results

See test report for an extended history of previous test failures. This is useful for diagnosing flaky tests.

       15 files  ±0         15 suites  ±0   6h 23m 23s ⏱️ + 13m 54s
  3 150 tests ±0    3 057 ✔️  - 2    86 💤 +1    7 +1 
23 307 runs  ±0  22 370 ✔️  - 2  915 💤 +1  22 +1 

For more details on these failures, see this check.

Results for commit e94b27e. ± Comparison against base commit 6afce9c.

@jrbourbeau
Copy link
Copy Markdown
Member Author

I'd prefer to keep that for a separate PR. I'm hoping we can make things compatible with bokeh=3 quickly and not need to set an upper bound. But if we can't then we can add the version constraint and message displayed in a follow-up PR

@jrbourbeau
Copy link
Copy Markdown
Member Author

Planning to merge in the next 48 hours if no further comment

@jrbourbeau jrbourbeau merged commit cd3554a into dask:main Oct 28, 2022
@jrbourbeau jrbourbeau deleted the fix-conda-install-bokeh branch October 28, 2022 15:02
mrocklin pushed a commit that referenced this pull request Oct 31, 2022
)

Closes #7227

- Prior to #7172 , the `bokeh` package was an optional dependency. (In the sense that you could start a cluster without bokeh installed)
  - In the most recent release (`2022.10.1`), `bokeh` became a required depdency due to the import of `distributed.dashboard.core` in `missing_bokeh` which raises an ImportError if bokeh is not installed. 
    - Not _strictly_ required, since the dashboard can be disabled by passing `dashboard_address=None` to the constructor of the Cluster. Which allows use of `distributed==2022.10.1` without bokeh installed. 

This PR moves the min bokeh variable to a different module that doesn't import code from the dashboard.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants