anonfile 1.0.0

[nstrydom2]

We are happy to announce the release of anonfile 1.0.0 🎉

This release includes updated dependencies and incorporates more robust testing. Check out our changlog for a more detailed list of changes. We recommend users upgrade to this version.

Please report any issues with the release on the anonfile issue tracker.

---

This discussion was created from the release anonfile 1.0.0.

[StefanGreve]

Regarding the recent build failures, I have looked a little bit into what's happening in the log files on GitHub Actions. Firstly, it is pretty clear that the error stems from the python -m build command (log).

In PR #95 I added the MANIFEST.in file hoping this would fix our problem. While that evidently doesn't seem to be the case, I still believe that this was at least a partial solution.

There is a closed issue on the build repository (pypa/build#427) that looks pretty similar to what we are encountering here. In the preceding issue (pypa/build#426) there's a comment that suggested the following fix:

python -m build --no-isolation

Apparently it takes some playing around to get the tooling to work correctly.

[StefanGreve]

@nstrydom2 I just remembered that rebuilding the job won't do anything good, because it will pull version 1.0.0 for the build since the current triggers are set to [published, edited]. In other words, to test the build fixes you need to create a new release with the tag v1.0.1 (which is actually in tune with what semantic versioning prescribes for this situation). I already went ahead and pushed this tag on master, but you will need to create a new release.

If this doesn't work out, we will need to retract release 1.0.1, delete tag v1.0.1 and repeat until we find a solution, though I am fairly confident that the next release should do the trick. When I run python -m build locally I get a successful build with and without the --no-isolation flag.

I added the workflow_dispatch event to the python-publish.yml workflow file as well. Version 1.0.0 is out now!

Obviously this situation was not ideal, but I have a few ideas how we can improve this workflow during development so that the actual release build doesn't need to be experimented with :)

Update: There's a feature proposal about this online now: #98