Speed up import and load hvplot lazily#1541
Merged
Merged
Conversation
Small change which reduces time for "import uxarray" on my machine from 1.7 seconds to 0.7 seconds. Also avoids the strange gray bars in Jupyter until actually making a plot for the first time, instead of immediately when importing uxarray. (The bars apparently came from "import hvplot.pandas" and "import hvplot.xarray"). See also: UXARRAY#1260 This also means hvplot import is lazy, which means we should be able to move hvplot to be an optional dependency instead of a requirement. See also: UXARRAY#1224
rajeeja
reviewed
Jun 30, 2026
Co-authored-by: Rajeev Jain <rajeeja@gmail.com>
Collaborator
Author
|
We might want to delay merging; I just discovered that import hvplot.pandas and import hvplot.xarray always change the backend to bokeh. Thus, with these changes, lazily importing hvplot means the first plot you make from uxarray will be forced into having bokeh backend, regardless of the backend you tried to set originally. I am creating a bug report on hvplot repository now about this. |
Collaborator
Author
|
I'm very confused about why my latest changes would cause those crashes… I only added an additional Any guesses on how my changes might have led to this? A quick attempt to reproduce it didn't lead me anywhere either, e.g.: uxds = ux.tutorial.open_dataset("outCSne30-vortex")
hasattr(uxds, 'plot')gives True. I can chase this down more if I should prioritize it, but wanted to check in first in case you have a guess already. Seems like it might be related to the |
6 tasks
hvplot has no Store attribute, so the previous guard always skipped the save/restore and the active backend was still forced to bokeh. Read/write the backend on holoviews.Store, and only restore backends that were actually loaded (Store.registry) to avoid breaking rendering on a fresh import.
11 tasks
Sevans711
added
the
improvement
ASV BenchmarkingBenchmark Comparison ResultsBenchmarks that have improved:
Benchmarks that have stayed the same:
|
cmdupuis3
approved these changes
Jul 9, 2026
cmdupuis3
left a comment
cmdupuis3
left a comment
Collaborator
There was a problem hiding this comment.
Wondering if there's a better way to collect all of the possible optional dependencies in one place, and import the accessors when need.
Doesn't seem necessary atm but would probably be good to use, like, lru_cache or something to force one-time loads across the whole package.
github-project-automation
Bot
moved this from 👀 In review
to ✅ Done
in UXarray Development
Collaborator
Author
|
Thank you for your review and comments, and merging to main! I think we shouldn't need to worry about making our own cache, because Python already handles import caching nicely. We just have to make sure optional dependencies and any "heavy"/"takes a long time to import" dependencies are only imported when we actually need them, rather than immediately upon import uxarray. Will continue optional dependency updates / discussion here: #1548 |
Contributor
|
Thanks for looking into this. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #1539
Overview
Speeds up
import uxarrayfrom ~1.7 seconds to ~0.7 seconds, by makinghvplot.pandasandhvplot.xarrayimports lazy, i.e. only import them when actually initializing a PlotAccessor object for the first time.PR Checklist
General
Testing