-
Notifications
You must be signed in to change notification settings - Fork 46
Run hindcast tests only on change to hindcast module #270
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
Conversation
@akeeste can I request a quick review on this PR to help us move PRs through review quicker due to less time spent on hindcast calls? |
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.
Hi Sterling, thanks for doing this. A couple minor comments included below
.github/workflows/main.yml
Outdated
@@ -211,6 +244,7 @@ jobs: | |||
coveralls: | |||
name: Indicate completion to coveralls.io | |||
needs: [conda-build, pip-build, hindcast-calls] | |||
if: always() && (needs.hindcast-calls.result == 'success' || needs.hindcast-calls.result == 'skipped') |
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.
Is this indicating that if the hindcast tests are run but fail, coveralls reporting is skipped? Will coveralls still consider the hindcast module to be covered if the hindcast tests don't need to run?
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.
If hindcast fails coveralls is skipped. You can see that as the current behavior (I grabbed the following screenshot from PR #268 which currently fails):
In my screenshot for this PR description I show the used logic. You can see that hindcast is skipped and that coveralls is submitted.
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.
What coveralls does is consider lines of code ran vs lines of code tested and aggregates that into a single number.
When we do not run hindcast the lines in the hindcast module do not count against coverage (in our case we will catch any degradation of coverage on hindcast in commits to master to when the hindcast module is changed).
Coverage is always a tenuous metric at best.
# MHKiT v0.8.0 We're excited to announce the release of MHKiT v0.8.0, which brings a host of new features, enhancements, and bug fixes across various modules, ensuring compatibility with Python 3.10 and 3.11, and introducing full xarray support for more flexible data handling. Significant updates in the Wave and DOLfYN modules improve functionality and extend capabilities. ## Python 3.10 & 3.11 Support MHKiT now supports python 3.10 and 3.11. Support for 3.12 will follow in the next minor update. - #240 ## Wave Module ### Enhancements: **Automatic Threshold Calculation for Peaks-Over-Threshold**: We've introduced a new feature that automatically calculates the "best" threshold for identifying significant wave events. This method, originally developed by Neary, V. S., et al. in their 2020 study, has now been translated from Matlab to Python, enhancing our existing peaks-over-threshold functionality. **Wave Heights Analysis**: A new function, `wave_heights`, has been added to extract the heights of individual waves from a time series. This function uses zero up-crossing analysis to accurately measure wave heights, improving upon our previous methods which only provided the maximum value between up-crossings. **Enhanced Zero Crossing Analysis**: Building on the above, the zero crossing code previously embedded in `global_peaks` has been isolated into a helper function. This modular approach not only refines the codebase but also supports new functionalities such as calculating wave heights, zero crossing periods, and identifying crests. ### Bug Fixes: **Contour Sampling Error in Wave Contours**: A bug identified in `mhkit.wave.contours.samples_contour` has been resolved. The issue occurred when period samples defined using the maximum period resulted in values outside the interpolation range of the contour data. This was corrected by ensuring that all sampling points are within the interpolation range and adjusting the contour data selection process accordingly. - #268 - #252 - #278 ## Xarray Support MHKiT functions now fully support the use of xarray for passing and returning data. - #279 - #282 - #285 - #302 - #310 ## DOLfYN Thanks to the many user contributions and users of MHKiT the DOLFYN module include a significant number of enhancements and bug fixes. ### Enhancements: **Altimeter Support**: Enhanced the Nortek Signature Reader to add capability for reading ADCP dual profile configurations. **Data Handling Improvements**: Introduced logic to skip messy header data that can accumulate during measurements collected via Nortek software on PCs and Macs. **Instrument Noise Subtraction**: Added a function to subtract instrument noise from turbulence intensity estimation using RMS calculations, providing results that differ by approximately 1% from the existing standard deviation-based "I" property. **Improved File Handling**: Updates for RDI files to handle changing "number of cells" and variable "cell sizes," which are now bin-averaged into the largest cell size. ### Bug Fixes: **Power Spectra Calculation**: Fixed a bug where a given noise value was not being subtracted from the power spectra, and noise was inadvertently added as an input to dissipation rate calculations. **Improved Header Handling**: Allowed RDI reader to skip junk headers effectively. **Nortek Reader C Types Update**: Adjusted C types in the Nortek reader to handle below-zero water temperatures and to allow pitch and roll values to go negative. - #280 - #289 - #290 - #292 - #293 - #294 - #299 ## River & Tidal: D3D Added limits to `variable_interpolation` and added 3 array input capability to `create_points` - #271 ## Developer Experience ### Black formatting Black formatting is now enforced on all MHKiT files. This ensures consistent formatting across the MHKiT package. - #281 ### Linting & Type Hints MHKiT is in the process of enforcing pylint and adding type hints to all functions. Currently this has been achieved and is enforced in the Loads and Power modules. - #288 - #296 ### CI/CD This release introduces significant reduction in testing time for development. This is achieved by reducing the number of tests for pulls against the develop branch and only running hindcast test when changes are made to it. A bug in the hindcast CI was fixed which only ran on changes to the hindcast tests instead of the hindcast module. Additionally the wave and wind hindcast needed to be separated in 2 jobs due to the excessive time taken to run a wind cache. This created a number of follow on PRs around solidifying the logic of these job. A special case for Python 3.8, pip, and Mac OS was added to use homebrew to install NetCDF and HDF5 to get tests to pass. - #241 - #270 - #306 - #311 - #317 - #318 - #319 - #320 - #324 ### Clean Up MHKiT fixed an implementation error where functions used assert instead of built in errors for type and value checking. Additionally these PRs removed unused files, fixed typos, and created an argument which allows users to run CDIP API calls silently. - #276 - #272 - #273
This Pull Request introduces a mechanism to selectively trigger the execution of expensive hindcast tests based on the context of the changes made. Specifically, the hindcast tests will only run under two scenarios:
mhkit/wave/io/hindcast
directory.master
branch.To achieve this, a new job named
check-changes
has been added to the workflow, which precedes theprepare-cache
job.Here's how the check-changes job operates:
mhkit/wave/io/hindcast
directory have been altered.wave_io_hindcast_changed
to true.wave_io_hindcast_changed
to false.wave_io_hindcast_changed
is true.should-run-hindcast
to true.should-run-hindcast
to false.Some additional minor modifications have been made to ensure the seamless integration of Coveralls reporting, regardless of whether the hindcast tests are executed or skipped.