@@ -33,30 +33,24 @@ Tag the release via::
3333
3434Then push the ``vX.X.X `` tag to the PyWavelets GitHub repo.
3535
36- Note that while Appveyor will build wheels for Windows, it is preferred to
37- get those wheels from the step below. Instructions for grabbing Appveyor
38- wheels manually here for reference only: if the commit with
39- ``ISRELEASED=True `` is submitted as a PR, the wheels can be downloaded from
40- Appveyor once it has run on the PR. They can be found under the "Artifacts"
41- tab in the Appveyor interface.
42-
43-
4436Build Windows, OS X and Linux wheels and upload to PyPI
4537-------------------------------------------------------
4638
47- Push a commit with the new tag and updates of dependency versions where needed
48- to https://github.com/MacPython/pywavelets-wheels. The wheels will be
49- produced automatically and uploaded to http://wheels.scipy.org/.
50- From there they can be uploaded to
51- `PyPI <https://pypi.python.org/pypi/PyWavelets >`_ automatically with
52- ``wheel-uploader ``.
39+ Pushing the vX.X.X tag to the repository will kick off automated build and
40+ deployment of the wheels to PyPI. The wheel builds proceed via GitHub Actions
41+ and their status can be checked by going to the Actions tab on GitHub.
5342
54- See the README on https://github.com/MacPython/pywavelets-wheels for more
55- details.
43+ In the event that the automated deployment fails, the built wheels can be
44+ downloaded via the GitHub Actions artifacts and then uploaded manually using
45+ twine as described below.
5646
5747Create the source distribution
5848------------------------------
5949
50+ The automated wheel build process should also automatically upload the sdist
51+ to PyPI. In the event that automated upload of the sdist fails, please proceed
52+ in generating and uploading it manually as described in this section.
53+
6054Remove untracked files and directories with ``git clean ``.
6155*Warning: this will delete files & directories that are not under version
6256control so you may want to do a dry run first by adding -n, so you can see what
@@ -76,8 +70,9 @@ Create the source distribution files via::
7670Upload the release to PyPI
7771--------------------------
7872
79- The binary Windows wheels downloaded from Appveyor (see above) should
80- also be placed into the ``/dist `` subfolder along with the sdist archives.
73+ These instructions cover how to upload wheels and source distributions to PyPI
74+ in the event that the automated deployment fails. The binary Windows wheels downloaded from GitHub Actions (see above) should also be placed into the
75+ ``/dist `` subfolder along with the sdist archives.
8176
8277The wheels and source distributions created above can all be securely uploaded
8378to pypi.python.org using twine::
@@ -92,8 +87,11 @@ documentation.
9287Update conda-forge
9388------------------
9489
95- Send a PR with the new version number and ``sha256 `` hash of the source release
96- to https://github.com/conda-forge/pywavelets-feedstock.
90+ The is an autotick bot run by conda-forge that is likely to autodetect the new
91+ PyPI release and autogenerate a PR for you that will update the PyWavelets
92+ feedstock for conda-forge. If this automated PR does not appear, you will need
93+ to send a PR with the new version number and ``sha256 `` hash of the source
94+ release to https://github.com/conda-forge/pywavelets-feedstock.
9795
9896
9997Create the release on GitHub
0 commit comments