Astral to be aquired by OpenAI

[snoopj@hachyderm.io]

@stylus I don't really expect anybody to make one, even in the probable case that OpenAI junk starts making its way downstream

`ruff` or `ty` have less of a combinatoric explosion from less "we're gonna do everything" task responsibility, but `uv` is doing so much that it seems to me that a squad's worth of people is necessary to maintain that feature-set.

[huronbikes@cyberplace.social]

@SnoopJ Interested to see the terms of the "purchase". If there's real cash being exchanged then this deal is the exit for the VCs who acquired Astral. If it's stock, then it's an attempt to further prevent the coyote from looking down.

[snoopj@hachyderm.io]

@huronbikes IMO this is absolutely the exit for those VCs but I don't know if we're ever going to get that kind of information

[huronbikes@cyberplace.social]

@SnoopJ Prior OpenAI acquisitions have mentioned the purchase method, but maybe we have reached the point in whatever this is where the purchase method is "don't worry about it."

[snoopj@hachyderm.io]

@huronbikes you're following it more closely than I am, so maybe I'm just wrong/naïve about that :)

[huronbikes@cyberplace.social]

@SnoopJ "following more closely" is me reading a thing then shitposting, so I could also be completely wrong about the details. Wouldn't read too much into it.

I'm just thinking of when OpenAI "bought" the guy who hated ports on Macbook Pros's company, which was supposedly valued as x billions of dollars but was done via OpenAI stock offerings.

[snoopj@hachyderm.io]

@huronbikes my rationale here that this is the off-ramp for those VCs is that surely they did not need Astral to get them in the door if they wanted in on OpenAI

that and that Astral's move with Pyx sure looked a lot like "the investors have ants in their pants for some returns"

but maybe it's something between the two extremes and this is one step towards paying out those VCs? as you suggest, we won't know until/unless we see some more details

[huronbikes@cyberplace.social]

@SnoopJ Yeah, that makes sense. I don't think that the VCs who bought Astral were looking for a specific someone, they just wanted a payday.

I hope it's not the case but ultimately this will probably lead to a degradation in the quality of the tools.

[swelljoe@mas.to]

@arrjay @SnoopJ the great thing about the Astral tools is they're not written in Python, so Python developers can't break it. Unlike every other tool that does the same things the Astral tools do, which break all the damned time.

I've never loved anything as much as Python developers love breaking backward compatibility, even for core functionality. Your CI worked yesterday and you think that means it'll work today? Don't be silly. pip, venv, pipenv, poetry, setuptools, etc. break all the time.

[swelljoe@mas.to]

@arrjay @SnoopJ the performance benefits are irrelevant, for me. I just love that every project I've converted to uv (always converted because something in the Python toolchain broke) hasn't had a broken build since I converted them.

[snoopj@hachyderm.io]

@swelljoe @arrjay I would make a case that `pipenv` and `poetry` are fundamentally broken tools, personally, but YMMV

hope that people who are standing where you are (that is, depending strongly on these tools) are not harmed too badly by the assorted insanity of the Money Bastards

[snoopj@hachyderm.io]

@swelljoe @arrjay actually, I would only have to make the case for Poetry because Chris Warrick already did an enviable job writing up the case for `pipenv`, saying everything I would have and more, with receipts:

https://chriswarrick.com/blog/2018/07/17/pipenv-promises-a-lot-delivers-very-little/

[swelljoe@mas.to]

@SnoopJ @arrjay I've said before that Poetry is king shit of failure in an ecosystem made of failure. Everything about packaging/deploying Python is poison, intentionally hostile to stability over time. I don't have strong feelings about Python, the language. It's fine...not my favorite, but it'll do. But, the ecosystem? The way Python devs view backward compatibility? Just hateful to people maintaining a lot of code that has to run on a variety of systems over many years.