-
Notifications
You must be signed in to change notification settings - Fork 297
Mesh tests #4034
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Mesh tests #4034
Conversation
* Add abstract cube summary (SciTools#3987) Co-authored-by: stephen.worsley <[email protected]> * add nox session conda list (SciTools#3990) * Added text to state the Python version used to build the docs. (SciTools#3989) * Added text to state the Python version used to build the docs. * Added footer template that includes the Python version used to build. * added new line * Review actions * added whatsnew * Iris py38 (SciTools#3976) * support for py38 * update CI and noxfile * enforce alphabetical xml element attribute order * full tests for py38 + fix docs-tests * add whatsnew entry * update doc-strings + review actions * Alternate xml handling routine (#29) * all xml tests pass for nox tests-3.8 * restored docstrings * move sort_xml_attrs * make sort_xml_attrs a classmethod * update sort_xml_attr doc-string Co-authored-by: Bill Little <[email protected]> * add jamesp to whatsnew + minor tweak Co-authored-by: James Penn <[email protected]> * normalise version to implicit development release number (SciTools#3991) * Gallery: update COP maps example (SciTools#3934) * update cop maps example * comment tweaks * minor comment tweak + whatsnew * reinstate whatsnew addition * remove duplicate whatsnew * don't support mpl v1.2 (SciTools#3941) * Cubesummary tidy (SciTools#3988) * Extra tests; fix for array attributes. * Docstring for CubeSummary, and remove some unused parts. * Fix section name capitalisation, in line with existing cube summary. * Handle array differences; quote strings in extras and if 'awkward'-printing. * Ensure scalar string coord 'content' prints on one line. * update intersphinx mapping and matplotlib urls (SciTools#4003) * update intersphinx mapping and matplotlib urls * use matplotlib intersphinx where possible * review actions * review actions * update readme badges (SciTools#4004) * update readme badges * pimp twitter badge * update readme logo img src and href (SciTools#4006) * update setuptools description (SciTools#4008) Co-authored-by: Patrick Peglar <[email protected]> Co-authored-by: stephen.worsley <[email protected]> Co-authored-by: tkknight <[email protected]> Co-authored-by: James Penn <[email protected]> Co-authored-by: Ruth Comer <[email protected]> Co-authored-by: Patrick Peglar <[email protected]> Co-authored-by: stephen.worsley <[email protected]> Co-authored-by: tkknight <[email protected]> Co-authored-by: James Penn <[email protected]> Co-authored-by: Ruth Comer <[email protected]>
* MeshMetadata class. * MeshMetadata extra members for dim names. * Comment for BaseMetadata refactoring.
* add mesh coordinate manager * wip * make shape methods private + reorganise method order * review actions * partial mesh * wip
* Update mesh-data-model branch (SciTools#4009) * Add abstract cube summary (SciTools#3987) Co-authored-by: stephen.worsley <[email protected]> * add nox session conda list (SciTools#3990) * Added text to state the Python version used to build the docs. (SciTools#3989) * Added text to state the Python version used to build the docs. * Added footer template that includes the Python version used to build. * added new line * Review actions * added whatsnew * Iris py38 (SciTools#3976) * support for py38 * update CI and noxfile * enforce alphabetical xml element attribute order * full tests for py38 + fix docs-tests * add whatsnew entry * update doc-strings + review actions * Alternate xml handling routine (#29) * all xml tests pass for nox tests-3.8 * restored docstrings * move sort_xml_attrs * make sort_xml_attrs a classmethod * update sort_xml_attr doc-string Co-authored-by: Bill Little <[email protected]> * add jamesp to whatsnew + minor tweak Co-authored-by: James Penn <[email protected]> * normalise version to implicit development release number (SciTools#3991) * Gallery: update COP maps example (SciTools#3934) * update cop maps example * comment tweaks * minor comment tweak + whatsnew * reinstate whatsnew addition * remove duplicate whatsnew * don't support mpl v1.2 (SciTools#3941) * Cubesummary tidy (SciTools#3988) * Extra tests; fix for array attributes. * Docstring for CubeSummary, and remove some unused parts. * Fix section name capitalisation, in line with existing cube summary. * Handle array differences; quote strings in extras and if 'awkward'-printing. * Ensure scalar string coord 'content' prints on one line. * update intersphinx mapping and matplotlib urls (SciTools#4003) * update intersphinx mapping and matplotlib urls * use matplotlib intersphinx where possible * review actions * review actions * update readme badges (SciTools#4004) * update readme badges * pimp twitter badge * update readme logo img src and href (SciTools#4006) * update setuptools description (SciTools#4008) Co-authored-by: Patrick Peglar <[email protected]> Co-authored-by: stephen.worsley <[email protected]> Co-authored-by: tkknight <[email protected]> Co-authored-by: James Penn <[email protected]> Co-authored-by: Ruth Comer <[email protected]> * Master to mesh data model (SciTools#4022) * Add abstract cube summary (SciTools#3987) Co-authored-by: stephen.worsley <[email protected]> * add nox session conda list (SciTools#3990) * Added text to state the Python version used to build the docs. (SciTools#3989) * Added text to state the Python version used to build the docs. * Added footer template that includes the Python version used to build. * added new line * Review actions * added whatsnew * Iris py38 (SciTools#3976) * support for py38 * update CI and noxfile * enforce alphabetical xml element attribute order * full tests for py38 + fix docs-tests * add whatsnew entry * update doc-strings + review actions * Alternate xml handling routine (#29) * all xml tests pass for nox tests-3.8 * restored docstrings * move sort_xml_attrs * make sort_xml_attrs a classmethod * update sort_xml_attr doc-string Co-authored-by: Bill Little <[email protected]> * add jamesp to whatsnew + minor tweak Co-authored-by: James Penn <[email protected]> * normalise version to implicit development release number (SciTools#3991) * Gallery: update COP maps example (SciTools#3934) * update cop maps example * comment tweaks * minor comment tweak + whatsnew * reinstate whatsnew addition * remove duplicate whatsnew * don't support mpl v1.2 (SciTools#3941) * Cubesummary tidy (SciTools#3988) * Extra tests; fix for array attributes. * Docstring for CubeSummary, and remove some unused parts. * Fix section name capitalisation, in line with existing cube summary. * Handle array differences; quote strings in extras and if 'awkward'-printing. * Ensure scalar string coord 'content' prints on one line. * update intersphinx mapping and matplotlib urls (SciTools#4003) * update intersphinx mapping and matplotlib urls * use matplotlib intersphinx where possible * review actions * review actions * update readme badges (SciTools#4004) * update readme badges * pimp twitter badge * update readme logo img src and href (SciTools#4006) * update setuptools description (SciTools#4008) * cirrus-ci compute credits (SciTools#4007) * update release process (SciTools#4010) * Stop using deprecated aliases of builtin types (SciTools#3997) * Stopped using deprecated aliases of builtin types. This is required to avoid warnings starting with NumPy 1.20.0. * Update lib/iris/tests/test_cell.py Co-authored-by: Bill Little <[email protected]> * Update lib/iris/tests/test_cell.py Co-authored-by: Bill Little <[email protected]> * Updated whatsnew. Co-authored-by: Bill Little <[email protected]> * celebrate first time iris contributors (SciTools#4013) * Docs unreleased banner (SciTools#3999) * baseline * removed debug comments * reverted * remove line * Testing * testing extensions * testing rtd_version * fixed if * removed line * tidy up * tidy comments * debug of pre-existing rtd variables * added reminder * testing * testing still * updated comments * added whatsnew * expanded the if conditiion * review actions * Update layout.html Remove alternative banner that used the RestructuredText notation. * review actions * drop __unicode__ method usage (SciTools#4018) * cirrus-ci conditional tasks (SciTools#4019) * cirrus-ci conditional tasks * use bc for bash arithmetic * revert back to sed * use expr * reword * minor documentation changes * review actions * make iris.common.metadata._hexdigest public (SciTools#4020) Co-authored-by: Patrick Peglar <[email protected]> Co-authored-by: stephen.worsley <[email protected]> Co-authored-by: tkknight <[email protected]> Co-authored-by: James Penn <[email protected]> Co-authored-by: Ruth Comer <[email protected]> Co-authored-by: Alexander Kuhn-Regnier <[email protected]> Co-authored-by: Patrick Peglar <[email protected]> Co-authored-by: stephen.worsley <[email protected]> Co-authored-by: tkknight <[email protected]> Co-authored-by: James Penn <[email protected]> Co-authored-by: Ruth Comer <[email protected]> Co-authored-by: Alexander Kuhn-Regnier <[email protected]>
* ConnectivityManager first pass. * ConnectivityManager align with proposed CoordManager. * Connectivity Manager review actions. * Connectivity Manager more review changes. * Use metadata_manager for Mesh location dimension. * Mesh dimension name abstraction. * Align Cooord and Connectivity Managers filters methods. * Completed Mesh class. * filter_cf improvements. * Moved filter_cf. * Mesh connectivity manager namedtuples comment. * Mesh removed trailing underscores. * Mesh _set_dimension_names improvements. * Mesh import rationalisation. * Mesh connectivity manager remove NDIM. * Connectivity manager use lazy indices_by_src(). * Connectivity manager clearer removal syntax. * Connectivity manager don't override __init__. * Connectivity manager correct base class syntax. * Metadata filter hexdigest reference fix. * test_MeshMetadata fix. * Rename filter to metadata_filter.
* minor fixes * wip
Something jazzy going on with the tests that I don't think can be related to my work. I'm trying running the whole suite locally to see what's up... |
… to play nicely with unittest.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@trexfeathers Great stuff!
Push the outcome of our offline conversation and I'll re-review again 👍
lib/iris/experimental/ugrid.py
Outdated
node = edge = face = False | ||
elif true_count == 0: | ||
# Treat None as True in this case. | ||
node, edge, face = [True if arg is None else arg for arg in args] |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@trexfeathers I kinda disagree with this change in behaviour.
What was there before was richer, where as what you're proposing now is quite restrictive and you're forcing the user to be explicit. Happy to chat about this to clarify.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Awesome thanks 👍
🚀 Pull Request
Description
Adds a suite of tests for the
Mesh
class.Test Style
In agreement with @bjlittle (and with tentative support from @pp-mo), these tests are written in a different style that is unit-esque, but focuses on being easier to write and read. You might call them "pragmatic tests" or some-such. The basic idea:
Cons
So an error takes more investigation.
But not much more, and errors always need some investigation anyway.
Pros
mock
, which is known to be difficult.Plenty of stories of
mock
making testing take longer than the functionality.I.e. more coverage for less thinking/writing.
With this style, I was able to write the
Mesh
tests - ~800 lines, ~80 tests, >100 assertions - pretty much fluidly without having to stop and work out how to write a test. This definitely hasn't been true in the past when trying to achieve proper unit test abstraction.The amount of time it takes to achieve good coverage with proper unit tests in my opinion often takes longer than the time they save when investigating errors.
This is of course only worth doing if it doesn't make the tests run excessively longer. For reference: testing the
Mesh
class - a fairly rich class - with reasonably good coverage takes ~100ms locally for me.If this style causes problems / doesn't solve problems in the months ahead, I'm happy to roll back on this idea. And by all means object if you have other concerns!
Consult Iris pull request check list