Talk:HarmonyOS

HarmonyOS identical to Android

Here are some interesting articles that could be used to update this Wikipedia page:

• No discernible difference between Huawei's "all-new" OS and Android , Ars Technica
• Huawei’s HarmonyOS appears to just be a forked version of Android, The Verge

Instanceof1 (talk) 16:05, 3 February 2021 (UTC)

It should be noted that these articles contain strong anti-Chinese politics speech, especially in comments. Other commenters have claimed and explained why these articles are fake, so it should be reviewed more clearly. 178.151.195.157 (talk) 12:11, 29 April 2021 (UTC)

The articles mentioned mainly focus on the technical aspects of the relationship between HarmonyOS and Android. Comments for or against the Chinese political system don't really matter. TeineiMudra (talk) 12:38, 8 May 2021 (UTC)

OpenHarmony is now open source on gitee, and HarmonyOS™ can be treated as a distribution of that. After comparing the structure and logic, it will be obvious that HarmonyOS does not resemble Android at all. 202.189.104.128 (talk) 16:29, 2 June 2021 (UTC)

Well, OpenHarmony has been open-source since 2019. The problem is that OpenHarmony cannot be built and run on Huawei phones, at least till now. 2A01:CB04:A63:DE00:1812:1C62:9D79:DFA3 (talk) 16:52, 2 June 2021 (UTC)

I would like to remind u all that HarmonyOS 1.0, HarmonyOS 2.0 and HarmonyOS 3.0 are totally different OS. just as Windows ME and Windows NT made also the same time were completely different OS. what seem to be clear is 2.0 does feature a layer of Android 10 to assure app compatibility, so there is at least a variant of android runtime in the OS. I believe the article should be spilt by the version since they are vastly different OSes with vastly different role and capabilities. 101.127.15.2 (talk) 00:42, 6 June 2021 (UTC)

From outside it looks like that THERE is no Android "layer" for Android app compability, but basically major parts of AOSP Android as it is. Ie, it is same kind of mod than CyanogenMod or OxygenOS or EMUI is. Only difference is that the Android texts and logos are removed. In theory there could be some kind of deeper changes like changing the hardware adaptation layer from Linux to another OS, but so far there is no anything which would prove that. Anyway, here is one explanation for whole HarmonyOS concept: https://www.youtube.com/watch?v=DLc-sp6FuWo Zache (talk) 09:29, 14 June 2021 (UTC)

The rebuttal article to the one from ArsTechnica is not in English, and (from what I can tell) the link is broken. It can't be used as evidence in this discussion. Unless a proper English version is available then it should probably be removed. Elahav (talk) 16:16, 2 December 2021 (UTC)

For clarity, while English language sources are preferred when there are sources of equal quality, there is no requirement that any sources must be in English and non English sources are perfectly acceptable, see WP:NOENG. However whatever language they are in, sources most qualify as WP:RS and a WP:SPS like a blog is rarely an RS. So Leo Li's popular science blog is probably not an RS and would not be even if it were in English. [1] As for a WP:dead link, well again provided there is sufficient information that the source can likely be retrieved in some manner, then it's generally still acceptable as a source. For a source which is only published online as is likely with a blog, it's trickier. Still at a minimum, a search should be made in common archival services and also in the host itself to see if it's just moved around. Nil Einne (talk) 15:25, 15 May 2022 (UTC)

It's still a mess, which is to be expected given the flips. The goalpost is always moving: you point out it's Linux, they say it's "new proprietary multi-kernel flexibility". You point out it's Java, they say it's "for familiarity and ease of development". You point out that Android .dex decompilers work fine, they pretend to not hear and keep repeating the last one. Don't even bring up the activity/ability rename -- it doesn't even make sense! --Artoria2e5 🌉 07:38, 19 April 2023 (UTC)

That is because it is more a project/framework than an "OS" in the traditional sense. You have it in the description. It can use multiple, vastly different, kernels, ONE of them being Linux. It support multiple, vastly different, application fameworks, again, ONE of them AOSP. Etc. Actually what makes HMOS a HMOS is most definitely the glue. The same like APT is what makes Debian be a Debian. Not the kernel, nor the DEs. The kernel abstraction layer, the modular concept with multiple side-by-side application framworks etc. In such a model the kernel, the framworks etc. are just components. Think of it like with Debian. It is a Linux distribution, except when it is not /Hurd/. It is an X11 system, except when it is not (Wayland or console-only). Etc. The fact most people commenting on it have some type of an agenda is not helping either.83.240.60.146 (talk) 23:06, 22 January 2024 (UTC)

Commons files used on this page or its Wikidata item have been nominated for speedy deletion

The following Wikimedia Commons files used on this page or its Wikidata item have been nominated for speedy deletion:

• HarmonyOS Service Center.jpg
• HarmonyOSCockpitUI.png

You can see the reasons for deletion at the file description pages linked above. —Community Tech bot (talk) 03:51, 16 November 2022 (UTC)

A Commons file used on this page or its Wikidata item has been nominated for speedy deletion

The following Wikimedia Commons file used on this page or its Wikidata item has been nominated for speedy deletion:

• HarmonyOS Connect IoT.png

You can see the reason for deletion at the file description page linked above. —Community Tech bot (talk) 07:06, 27 November 2022 (UTC)

Commons files used on this page or its Wikidata item have been nominated for speedy deletion

The following Wikimedia Commons files used on this page or its Wikidata item have been nominated for speedy deletion:

• AITOM5M7.png
• HarmonyOS3HomeNew.jpg

You can see the reasons for deletion at the file description pages linked above. —Community Tech bot (talk) 07:51, 6 December 2022 (UTC)

Misuse of the term "multikernel"?

This article, and a number of other articles that discuss or even just mention HarmonyOS or OpenHarmony, use the term "multikernel", and link to multikernel.

The multikernel article is a stub that says:

A multikernel operating system treats a multi-core machine as a network of independent cores, as if it were a distributed system. It does not assume shared memory but rather implements inter-process communications as message-passing. Barrelfish was the first operating system to be described as a multikernel.

This does not mean a "design with dual frameworks: the operating system selects suitable kernels from the abstraction layer in the case of devices that use diverse resources", as the lead paragraph says about HarmonyOS's "multikernel"; there's nothing in multikernel about being able to select a particular kernel from a set of kernels.

From what I can tell from the diagram in HarmonyOS § HarmonyOS NEXT, OpenHarmony and, presumably, HarmonyOS have a "kernel abstraction layer" atop which either the Linux kernel or the LiteOS microkernel can run, and HarmonyOS NEXT runs some unspecified microkernel (LiteOS-derived?) atop the kernel abstraction layer. That's nothing like a multikernel in the sense of that article; perhaps the article should speak of "what Huawei calls a 'multikernel'", and not link to multikernel.

In addition, the article should discuss the details of that (non-multikernel) architecture in HarmonyOS § Architecture, preferably with more technical detail (please let some Huawei engineers write it without the interference of the marketing department; I no more want to read what Huawei's marketing department has to say about the architecture of HarmonyOS than I want to hear what Apple's marketing department says about the architecture of macOS or iOS etc.. or what Microsoft's marketing department has to say about the architecture of Windows NT). Guy Harris (talk) 07:35, 1 May 2024 (UTC)

If that's the case, explain why Distributed operating system in the "See also" page. Poppodoms (talk) 08:42, 1 May 2024 (UTC)

alongside within Distributed operating system wiki page. Poppodoms (talk) 08:44, 1 May 2024 (UTC)

A distributed operating system is not necessarily a multikernel operating system. A multikernel operating system is an OS that runs on multi-core systems and that treats individual cores as if they were, to quote the first paragraph of distributed operating system, "physically separate computational nodes", and thus treating the multi-core system as a loosely-coupled multiprocessor rather than a tightly-coupled multiprcessor. All of the various Huawei OSes in question may well function as distributed operating systems across multiple separate machines even if they function as Boring Old SMP systems on a particular machine, but they're not multikernel unless they do the same sort of thing that Barrelfish does. They might be "multi-kernel" if they support putting more than one kernel atop the sadly-undescribed "kernel abstraction layer", but that's not necessarily "multikernel" in the multikernel sense. Guy Harris (talk) 08:56, 1 May 2024 (UTC)

And, in fact, HarmonyOS § Architecture says

The system includes a communication base called DSoftBus for integrating physically separate devices into a virtual Super Device, allowing one device to control others and sharing data among devices with distributed communication capabilities.

If by "physically separate devices" they truly mean devices that don't, for example, physically share memory with each other, unlike the cores on a multi-core system, with the communication going over a network, then that truly is a distributed operating system, but, again, it's only a multikernel system if, even the not-physically-separate cores on a multi-core CPU interact with each other via message passing. Guy Harris (talk) 09:01, 1 May 2024 (UTC)

Thanks for the clarification. Your explanation was very clear. :) Poppodoms (talk) 09:05, 1 May 2024 (UTC)

References