Python Software Foundation Is Hiring a Python Packaging Project Manager | Hacker News Reader

Python Software Foundation Is Hiring a Python Packaging Project Manager

pyfound.blogspot.com

90 pointsrbanffy38 comments

Hi, PSF director and PyPI maintainer here! Happy to answer any questions about this role.

This joins our other open position for a CPython Developer-in-Residence: https://pyfound.blogspot.com/2021/04/the-psf-is-hiring-devel...

Have you talked to the author(s) of Poetry about the job? Or at least a collab?

As far as I'm aware, we haven't talked to anyone specific about the role, but they're welcome to apply, of course.

Note that this is for a full-time project and community manager, not a developer (although there may be some overlap).

Yes this is definitely a good idea.

What is the general plan of the project, updating abs standardizing the current system or building a new package manager

Continued maintenance and new feature development for the existing ecosystem.

Shouldn't you simply be hiring Donald Stufft for this position?

Donald already has a full time job!

Also I'm not sure how much he'd enjoy being a project manager. :)

Is there any plan to deal with non python dependencies somehow - in a way that's consistent?

E.g. install psycopg2 <- requires apt-get install libpq (and that's just on Debian) but python packaging has no consistent way of managing this.

We don't currently have any plans to address this issue specifically -- this is a good use-case for using a general package manager like conda rather than something Python-specific like pip.

That's disappointing to hear. I had suspected it was consigned to the realm of "somebody else's problem" but it's sad to hear it confirmed officially.

I don't think any language actually solves this, and to be honest I'm in favour of the "it's someone else's problem" approach. Python packaging is complex enough without having Python try to actually solve this massively complicated secondary problem.

I'm happy to see this. The python packaging ecosystem is frustratingly nonstandardized, particularly the split between conda and pip. Now that the Python 2->3 migration has mostly been dealt with, hopefully the community is able to improve the packaging situation.

Could not agree more

Python with one computer = Pure Joy

Python shared to two or more computers utterly devastating. I left python a long time ago when I couldn't get my code to easily repeat on three of my computers at three locations for work. I also tried a simple program to use on 100 computers at 40 different locations was a disaster (Luckily no one knew)

I actually used Racket and it was a the easiest packaging system I ever used and had smooth sailing on projects that I used for pet projects that saved me tons of time.

This is pretty exciting. There's a lot of cruft there that nobody dealt with - often because nobody really saw it as their job to do so.

This is fantastic. Wholly support non-profit funded employees working to maintain system critical open source software like Python.

Is it fair to call this non-profit funded? The money comes from Bloomberg and Google.

The PSF, which is the organization hiring for this job, is a non-profit. As with virtually all non-profits, the PSF is funded largely by donations.

The Steering Council is Microsoft, Google, Bloomberg and has already stifled free development and opinions.

I expect that corporations will get even more power with full time employees and very little will be accomplished.

Can the python packaging story get much worse? It's a huge blemish and way too much fragmentation and after market solutions, and of course the mental overhead of virtualenvs or else you clobber your environment.

One thing I keep hoping will get advanced in is PEP-582 to incorporate a local __pymodules__ folder since it would be nice if getting a python project up and running was as easy as node or php.

> Can the python packaging story get much worse?

Of course it can get much worse, see where Python packaging was ~5 years ago.

You mean they will stiffle all those wonderful things accomplished the last 30 years in the Python Packaging ecosystem? /s

> has already stifled free development and opinions.

This is a very strong statement. Do you have any details about what you're claiming or supporting evidence?

Lol someone taking a little more power with Python packaging would be nice

About damn time!!

This is long overdue.

I have always envisaged that it would only happen if a corp steps in and wished I had the means.

Kudos to Bloomberg for stepping up and hope other corps add to the effort for actual programming

Is the need for a project manager or a product manager?

I’m pretty sure that python is going to have an ongoing need for attention to be paid to its packaging ecosystem. Approached holistically, this would include the documentation and other aspects of developer experience.

Hopefully the role prioritizes blessing existing good practices and creating migration paths and decent interop vs creating new things and pulling an xkcd 927.

I know everyone's got their favourite thing to be fed up about in the Python packaging ecosystem, but for me it's definitely `setup.cfg` vs `pyproject.toml`.

Have you seen PEP 621? It sets a standard for putting metadata in `pyproject.toml`.

https://www.python.org/dev/peps/pep-0621/

I'm well aware. Have you seen this drama about it?

https://github.com/python/mypy/issues/5205#issuecomment-4102...

That comment was made two years before PEP 621 even existed.

Okay, so then maybe my issue is that four years existed between PEP 518 and 621. :|

There really isn't any "drama" in that discussion. pyproject.toml is the way forward and projects will need to adopt it at some point.

Provided both are supported while new projects move to pyproject.toml

I really don't like how projects are being broken

Happy to see an alternative to json and YAML being used. TOML has been a delight whenever I've used it.

Scratching my head, would they do that without a toml parser in std?

thank fuck for that. There’s a lot of good stuff being done and a cat-herder with funding is desperately needed to keep it moving.

is there any good reference article/blog that talks about various python package managers, their history, use cases and the current trends in industry? Is there any specific tool that's becoming more accepted standard among developers?

Reply on news.ycombinator.com