Use reusable GitHub CI workflows for `test-on-pr` and `matrix-test` once Gooey package Py313 is available on Conda-forge

[bobleesj]

Problem

The test-on-pr and matrix-test CI .yml files in this repository have been modified to pip install all pip.txt dependencies to support Gooey while testing. Please see the PR here #134

Proposed solution

Once Gooey provides conda packages for Python 3.13, update the test-on-pr and matrix-test CI files using the Billinge group's reusable CI scripts.

[cadenmyers13]

@sbillinge we should move this to next release. Getting Gooey up to 3.13 could be a heavier lift and we want a good working version of labpdfproc out soon because of the paper publish

[sbillinge]

before I do this, could we check. I think we can make this pass if we remove gooey from conda.txt (it is probably already not there) but then add a line that explicitly pip installs gooey. I think this is working on regolith for example. I think that gooey actually works with these more recent pythons too, just can't be conda installed.

[cadenmyers13]

I think we can make this pass if we remove gooey from conda.txt (it is probably already not there) but then add a line that explicitly pip installs gooey.

@sbillinge yeah that is how it currently operates (see pasted below). The reusable CI does the testing via conda install. We can't move to the reusable workflow unitl gooey is on the latest conda, which is the problem that this issue is addressing.

tests-on-pr.yml in labpdfproc (custom)

name: Tests on PR

on:
 push:
 branches:
 - main
 pull_request:
 workflow_dispatch:

jobs:
 validate:
 defaults:
 run:
 shell: bash -l {0}

 runs-on: ubuntu-latest
 steps:
 - name: Check out diffpy.labpdfproc repository
 uses: actions/checkout@v4

 - name: Initialize miniconda
 uses: conda-incubator/setup-miniconda@v3
 with:
 activate-environment: test
 auto-update-conda: true
 environment-file: environment.yml
 auto-activate-base: false
 python-version: 3.13

 - name: Conda config
 run: >-
 conda config --set always_yes yes
 --set changeps1 no

 - name: Install diffpy.labpdfproc and requirements
 run: |
 conda install --file requirements/tests.txt
 conda install --file requirements/conda.txt
 pip install gooey
 python -m pip install . --no-deps

 - name: Validate diffpy.labpdfproc
 run: |
 pytest --cov
 coverage report -m
 codecov

 - name: Upload coverage to Codecov
 uses: codecov/codecov-action@v4
 with:
 verbose: true
 fail_ci_if_error: true
 token: ${{ secrets.CODECOV_TOKEN }}

tests-on-pr.yml in srfit, for example (general, uses reusable workflow)

name: Tests on PR

on:
 pull_request:
 workflow_dispatch:

jobs:
 tests-on-pr:
 uses: scikit-package/release-scripts/.github/workflows/_tests-on-pr.yml@v0
 with:
 project: diffpy.srfit
 c_extension: false
 headless: false
 secrets:
 CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}

[sbillinge]

Here is how the tests_on_pr workflow handles it in regolith:

name: Tests on PR

on:
 pull_request:
 workflow_dispatch:

jobs:
 tests-on-pr:
 uses: scikit-package/release-scripts/.github/workflows/_tests-on-pr.yml@v0
 with:
 project: regolith
 c_extension: false
 headless: false
 python_version: 3.13
 run: pip install gooey
 secrets:
 CODECOV_TOKEN: ${{ secrets.CODECOV_TOKEN }}

(note the line third from bottom)

[sbillinge]

regolith needs to be updated to latest skpg so the python version is auto-detected, but I think the run will still work and install the gooey. Make sure it is not present in the conda.txt. We will need an issue to fix this ugly workaround when we fix gooey, but I think this works for now.

[cadenmyers13]

@sbillinge Ah, i see. Sorry, I thought it all had to be manually done. I'll give that a shot.