undefined | Better HN

0 pointsjauntywundrkind2y ago0 comments

I was referring more to the harmony ecosystem, including OpenHarmony, which does have a variety of uses.

I agree it's too early to have strong conclusions. But there's a huge range of sectors represented at a recent-ish OpenHarmony conference. Industrial, energy, vehicle uses listed. https://mp.weixin.qq.com/s/eGxegRJCKhuYz8IvsxhE5g

The use of a variety of different kernels & microkernels, bound by a common DBusSoft distributed layer, is what seems compelling to me. It means one can have very very small microcontrollers readily participating in Harmony, and ideally it means interesting Plan 9 like capabilities across bigger computing (phones/desktop/tvs/&c), where inter-system control just works. It could all go nowhere or fizzle... but I see very narrow specific heavily industrialized works like Matter, which is trying desperately just to make multiple controllers capable of running one device (an incredibly low bar for ubicomp), and I am glad & excited someone else is working to connect computing systems like Harmony is.

The technicals have some compelling ideas.

0 comments

manuelabeledo2y ago

> I see very narrow specific heavily industrialized works like Matter, which is trying desperately just to make multiple controllers capable of running one device

There is not a lot of money in interoperability, so this is not an issue with the technology in itself, but the manufacturers will to implement it.

Distributed embedded computing isn't new at all, and one can run any sorts of fancy software on an ESP32 if need be, e.g. AtomVM.

HarmonyOS is a shot at technology independence, which is a fine goal in itself, but the chances of it being successful in Europe or the US are minimal at best.

jauntywundrkindOP2y ago

> There is not a lot of money in interoperability, so this is not an issue with the technology in itself, but the manufacturers will to implement it.

Based only off how utterly unwilling & uninterested the market seems to be, yeah, I am left with a similar feeling that computing is being hosed by exactly this misaligned incentives.

That Matter, which seemed like an honest effort at damage control to right this, has struggled so mightily (most glaringly when having multiple-controllers for a device), well, that could indicate a lot of things. But there being such a mess of ideas and implementations, and trying to groom it all latter/now, that seems like a foreseeable issue that having a better designed up front interoperable distributed system could have remedied.

DBusSoft to me is this an exciting possibility, trying bolder steps than anything that's been tried since Plan 9. You show how limited & unwilling to consider this are in one word of your dismissal:

> Distributed embedded computing isn't new at all

Sure there's plenty of attempts at distributed embedded systems. One can't help but run into dozens upon dozens of mqtt based schemes or others.

But this isnt a distributed embedded OS. Harmony's distinct ambition, what makes it a new Plan 9 like, is that DSoftBus underpins everything. It underpins embedded devices, the consumer OS, and apps. It harmonizes computing, be those local or far bits of computing; that's the dream at least.

I'm far from confident they'll succeed with Harmony. But I am elated & thrilled to see some deep systematic ideas under the hood of an ambitious cross-modal ubiquitous & pervasive OS.

Who knows how this goes, but the apathy & unwillingness of those who see "not a lot of money on interoperability" has me tired as hell, and it feels like we are stuck with shitty bad mainframes everywhere that the internet dials us into; I am desperate for more, and Harmony has least has some pretense of being a better connected computing. And there's little more I care about than that. Making general purpose computing participatively connected is I think what it takes to awake from this slumber of apathetic cloud computing & make computing compelling & personal again.

j / k navigate · click thread line to collapse