You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
All PRs/issues attached to the release are merged.
All the badges on the README are passing.
License information is verified as correct. If you are unsure, please comment below.
Locally rendered documentation contains all appropriate pages, including API references (check no modules are
missing), tutorials, and other human-written text is up-to-date with any changes in the code.
Installation instructions in the README, documentation, and the website are updated.
Successfully run any tutorial examples or do functional testing with the latest Python version.
Grammar and writing quality are checked (no typos).
Install pip install build twine, run python -m build and twine check dist/* to ensure that the package can be built and is correctly formatted for PyPI release.
Please tag the maintainer (e.g., @username) in the comment here when you are ready for the PyPI/GitHub release. Include any additional comments necessary, such as version information and details about the pre-release here:
PyPI/GitHub full-release preparation checklist:
Create a new conda environment and install the rc from PyPI (pip install <package-name>==??)
License information on PyPI is correct.
Docs are deployed successfully to https://<github-username-or-orgname>/<package-name>.
Successfully run all tests, tutorial examples or do functional testing.
Please let the maintainer know that all checks are done and the package is ready for full release.
conda-forge release preparation checklist:
Ensure that the full release has appeared on PyPI successfully.
New package dependencies listed in conda.txt and tests.txt are added to meta.yaml in the feedstock.
Close any open issues on the feedstock. Reach out to the maintainer if you have questions.
Tag the maintainer for conda-forge release.
Post-release checklist
Run tutorial examples and conduct functional testing using the installation guide in the README. Attach screenshots/results as comments.
Documentation (README, tutorials, API references, and websites) is deployed without broken links or missing figures.
@john-halloran I pushed out 0.3.0-rc.0. Please can you test the install in a 3.14 env and check the tests and all examples run successfully. Check the docs, confirm all is ok with me and I will complete the checklist and push out the full 0.3.0 release.
The README is showing that it is not on conda-forge. Did we never get it onto CF?
@john-halloran I pushed out 0.3.0-rc.0. Please can you test the install in a 3.14 env and check the tests and all examples run successfully. Check the docs, confirm all is ok with me and I will complete the checklist and push out the full 0.3.0 release.
The README is showing that it is not on conda-forge. Did we never get it onto CF?
No, because we needed another release to make it possible to be on CF. This should be the first time we can, and I'm going to do extra testing during the rc process to make sure it'll work this time. Thanks for jumping on this.
Observations from 0.3rc testing: Issues with docs:
Some API information is missing for newly added functions. This is in the release checklist and I’m happy to fix, but this could also be put on hold as docs improvments are a 0.4 issue
Notes section is scrambled (0.2 appears above title and does not render all its bullet points correctly. Other versions are below the title, which incorrecly says “otes” rather than “Notes”. We may prefer “Versions” or “Patch Notes” here)
No release notes are listed for 0.3 at all (but I recall this is standard for an rc)
Issues with installation:
We say to install with conda, but that is not currently possible. Will be fixed post release. Installing with pip on 3.14 worked fine
Issues with program operation:
“snmf —version” returns “snmf 0.0.1” instead of the expected “diffpy.stretched-nmf 0.3rc” I believe this may again be partly an RC issue.
There were no issues with actual program operation, including the example and runs on other data
Please can you check all the workflows. Are they the scikit package latest? When I tried to run the matrix workflow I get this:
This one is slightly out of date, but the fix is small. Are these workflows meant to be auto-updated when scikit-package is? Because we are already on the latest version of that. To fix this particular issue, all that needs to be done is remove the release_branch: main line (line 20) from matrix-and-codecov-on-merge-to-main.yml. That input has been deprecated upstream.
Will this prevent a conda-forge release, or is it just something we'd like to get fixed for 0.4?
I think it won't affect CF so it can go in the next release, but it will show up as a failing badge in the README so better to fix it sooner rather than later. I strongly recommend doing a quick update and making sure no other workflows are not the latest.
@sbillinge We are officially on conda, so is this safe to close? My only observations are that online docs are out of date, but I think we knew that. It might also be nice to have a link on the diffpy.org "other packages" page.
follow instructions to do a clean install from conda forge and run all tests and examples. Confirm all is well.
I think we have issues for the docs so nothing to do there
update the skpkg. It was done recently but the fact that the matrix build is failing raises uncertainty. We can remove this by doing the update. We don't need a new release, it will just go out with the next one. Confirm it's done
make issues for anything else you found. Confirm.
plan and execute a rollout (advertising this new development). This may or may not include a new short paper (I think yes)
Then we are done. We could include Filippo in that paper but only if it makes sense and he is contributing examples. If he has examples we can include them in the docs and the paper.
Activity
sbillinge commentedon Aug 15, 2026
@john-halloran I pushed out
0.3.0-rc.0. Please can you test the install in a 3.14 env and check the tests and all examples run successfully. Check the docs, confirm all is ok with me and I will complete the checklist and push out the full0.3.0release.The README is showing that it is not on conda-forge. Did we never get it onto CF?
john-halloran commentedon Aug 15, 2026
No, because we needed another release to make it possible to be on CF. This should be the first time we can, and I'm going to do extra testing during the rc process to make sure it'll work this time. Thanks for jumping on this.
john-halloran commentedon Aug 17, 2026
@sbillinge
Observations from 0.3rc testing:
Issues with docs:
Issues with installation:
Issues with program operation:
sbillinge commentedon Aug 17, 2026
I pushed out the full release. Please keep note of the documentation issues and make sure to fix them as part of the
0.4.0worksbillinge commentedon Aug 17, 2026
Please can you check all the workflows. Are they the scikit package latest? When I tried to run the
matrixworkflow I get this:john-halloran commentedon Aug 17, 2026
This one is slightly out of date, but the fix is small. Are these workflows meant to be auto-updated when
scikit-packageis? Because we are already on the latest version of that. To fix this particular issue, all that needs to be done is remove therelease_branch: mainline (line 20) frommatrix-and-codecov-on-merge-to-main.yml. That input has been deprecated upstream.Will this prevent a conda-forge release, or is it just something we'd like to get fixed for 0.4?
sbillinge commentedon Aug 17, 2026
I think it won't affect CF so it can go in the next release, but it will show up as a failing badge in the README so better to fix it sooner rather than later. I strongly recommend doing a quick update and making sure no other workflows are not the latest.
john-halloran commentedon Aug 18, 2026
@sbillinge We are officially on conda, so is this safe to close? My only observations are that online docs are out of date, but I think we knew that. It might also be nice to have a link on the diffpy.org "other packages" page.
sbillinge commentedon Aug 18, 2026
Congrats! HUGE accomplishment!
Please:
Then we are done. We could include Filippo in that paper but only if it makes sense and he is contributing examples. If he has examples we can include them in the docs and the paper.