Official upstreaming efforts are a significant investment even if they have internal PoCs. You can't just "release what you've got".

Sure you can. That said, I'm not sure what Apple's usual style is and it might be pretty out of character (or not) to drop support shortly after an MR.

No you can't. Upstreaming a large feature like this requires at the very least allocating a number of manweeks to the upstreaming process, and that's without accounting for actually writing the feature in the first place. Making a PR and not following up would be worse PR (heh) than doing nothing, which is also costly.

If they had dumped a tarball or a fork somewhere with a hacked up PoC it could have looked like they just dumped what they had, but even that requires approvals and time from legal, and so someone has to decide its worth investing in.

(I have no interest in speculating about the Vision lineup itself, just commentating on open source contributions.)

You really don't understand how much money Apple's sunk on this and how little comparatively it takes to wind down operations in a way that makes partners and suppliers happy. Keeping a couple small teams going, even with entirely new versions shipping, to keep your supply chain whole and your partners coming back for future products is cheap and easy and very likely what Apple's doing now that most of the top talent was moved to spectacles and Siri. Letting your OS and partner software teams clear their pipelines is absolutely common at Apple and elsewhere. Denying that is silly.

Nothing about what you said binds Apple to following through. Just because it would be a bad idea doesn't mean they can't or won't.

They could also throw money out their window, light it on fire, let their employees do whatever they want instead of regular Apple work, etc.

A lot of managers (and legal) have to approve and allocate significant budget to bad ideas. Companies generally try to avoid doing bad ideas that are a net negative.

Not if it was 90% done and you want to clear the decks of nearly completed work before moving on. It's absolutely common to ship stuff to make third party suppliers and partners whole even when you've decided not to continue investing in a product. Keeping partners happy is critical to the smooth functioning supply chain Cook built and which gives Apple most of their success today.

A big feature like this will be at most 20-30% done when you open a PR, because the rest of the work is in collaboration to suit upstream which will as always require notable changes and rewrites.

Spending manweeks to manmonths on new upstream work that won't see any practical use until half to a whole year later, after getting legal to work things out, is a significant investment in new development.

This has nothing to do with suppliers or partners. They are not dumb enough to measure a product roadmap in number of random PRs, but will look at first party software and hardware roadmap.

This is precisely what they're doing, shipping everything that was mostly already cooked, clearing out the pipleline for their own developers and for third parties, particularly the supply chain they left high and dry when they ended Vision Pro production early, but also their own OS devs and the ones working with third party software and content sponsorships.

They've already moved all the top talent and the next thing looks like a PCVR device that guts the PC from Vision Pro and maybe a chip and ship bump or even a cheaper model with sub-premium materials and lower fidelity optical stack.

What ever is going on, it's clearly not the priority it was and most people paying close attention see a steep decline in the viability of goggles at Apple. That form factor was a flop, as the ergonomics simply didn't align with the use cases for most normal people.