Why ?

[dfurtado]

Hi there,

I'm the maintainer of the dataclass-csv.

Great that you like my work, I appreciate it. However, wouldn't be better to add new features and fixes to the original project?

This project is active, and it is used by quite a lot of people, and it is confusing to have a dataclass-csv2 in PyPI, isn't it?

This is not what open-source is about.

Please, consider removing the additional package from PyPI and I'm open to work together with you or anyone else. There's a lot of improvements I wanted to in the CI/CD pipeline, documentation and new features to make the dataclass-csv much more flexible, and it would be nice to have someone to work with.

I worked really hard building this project and I got help from a few contributors during the years, and it is not fair to us doing something like this.

//Daniel

[kraktus]

Hey! Thanks for reaching out. I created the fork because the issue I was encountering dfurtado/dataclass-csv#52 was opened for more than three years at that point without any progress.

As well as some other PRs that were useful to me and had been opened for more than a year dfurtado/dataclass-csv#57 and dfurtado/dataclass-csv#61.

This is not a criticism btw, I'm the maintainer of another python project https://github.com/lichess-org/berserk and I know too well it's sometimes cumbersome to review them :p

It proved to be a useful test-ground for migrating to uv for package publishing. I was not sure it would be met positively upstream. If you want I can help you maintaining dataclass-csv, but I have rather sparse free-time nowadays.

[dfurtado]

Hi, thanks for the quick response.

The dfurtado/dataclass-csv#52 I just merged a PR with the fix and I'll be happy to accept the other PRs with different fixes for sure.

I've been trying to do too much at the same time and delayed quite a bit the issue response but I'm trying to catch up and tackle all of them now

At the moment, I'm currently working on CI/CD, and doing everything using Github actions instead, that includes, testing, linting, type annotations validation, and publishing package to PyPI

When it comes to test I wanted to implement a matrix that run the tests with different versions of Python. Since I have so many things planned it is a nice thing to ensure that the project is backwards compatible with older versions.

Regarding the uv it seems interesting, that's something that could be nice to add to the project.

Already worked on:

• Added a fix for the issue 52
• removed pipenv
• removed travis and use github actions for running the tests
• Just approved the issue 57 as well.

Currently working on:

• Run tests with different versions of python (tox)
• Add type annotation linter
• Going through the issue list

[kraktus]

I would honestly recommend to base your work on this fork which has some very up-to-date tooling, you’ll save a bunch of time.

[dfurtado]

Absolutely, I will check it out for sure.

You could open a issue on the main project and describe the changes you did in the fork and if you want you're very welcome to send me some PRs that way I don't need to just copy and paste the work you have done, you can add the PR and I can give you the credits by adding you in the contributors.

After we merge your PRs I would really appreciate if you unpublish the dataclass-csv2 because I've got messages from a couple of people confused which package they should install. (that's how I've got to know the new package existed)

Please, open an issue and let's discuss your changes, I would be happy to include the changes in the original project

[kraktus]

Hello, I've opened a PR for you to see the changes dfurtado/dataclass-csv#66, it has probably many merge conflicts I will not handle, but I'd suggest to take the code related to uv for building and publishing the package. Cheers.