Hacker News Reader

Comment by coldtea | Hacker News Reader

Give it some time, it will be mostly just on the server again -- as Android will be switching to Google's own OS:

http://www.theverge.com/2016/8/15/12480566/google-fuchsia-ne...

That is complete speculation. There are tons of reasons why Google wouldn't switch Android "over", including established development environments, drivers, and licensing.

It's fun to speculate on the purpose of this but it's just that: speculation.

Don't get me wrong: there are massive improvements to the existing linux ecosystem just ripe to be made, but many of them are in the environment and not just the kernel.

And by the same coin there are a lot of reasons they would.

Drivers can be fork lifted by implementing KPI compatibility, and that can be done in a shortest path fashion or done in a way to realize new protection from the new OS. Development environments can be preserved indefinitely with image activation and personalities. The first never mattered to Apple, and the second was effectively realized on OS X.

Not by the same coin, I'm saying that it's speculation. And, I provided some examples as to why it's speculation.

There is no reason to believe that Linux, and other possibly OSs, won't all grow during this time. As long no one is monopolistic and anti-competitive there is a ton of room for competition in this space.

The network effect, the winner takes all effect, is becoming less pronounced as there are more ways to build cross platform software.

>There is no reason to believe that Linux, and other possibly OSs, won't all grow during this time. As long no one is monopolistic and anti-competitive there is a ton of room for competition in this space.

Well, grow enough to counter-balance the sudden loss of billions of Android devices? I don't think there's reason to believe that. It had stalled for ages on the desktop, and mostly got to devices outside the embedded space through Android.

(Of course Android-Linux devices will continue to exist -- but eventually those with the new Google kernel/OS underneath will overtake them, especially if Google insists on OEMs using it to get access to the various Google apps).

>The network effect, the winner takes all effect, is becoming less pronounced as there are more ways to build cross platform software.

It's not about having some cross platform software though, it's about the whole ecosystem. Which is e.g. one of the reasons why Firefox OS, et al never got anywhere.

The ecosystem is important. What do you think about systems that make it easy to ship software for multiple platforms?

Consider Unity, the game making tools. When shipping there is pretty much a series of checkboxes for what kinds of packages you want to make. Unless you went out of your way to use some OS specific stuff it will work everywhere. There are other tools like this for other non-gaming domains, do they contribute in your view of OS ecosystems?

There is reason to believe that. Linux hasn't fundamentally changed in .. well has it ever? The driver model still sucks (I guess this is probably the main reason Google is moving away from it), the security is still embarrassing, the graphics story is very much stuck in the 90s (does display hot-plugging work yet? what about secure password entry?). It started as a Unix clone and it's still a Unix clone. What makes you think that will change?

I think the time is ripe for a new OS. Especially given that Google's OSes don't expose much of the OS to apps - it should make it quite easy to port their userspaces over.

I think that many of you complaints are without merit.

For fundamental there have been several schedulers, the removal of the "Big Kernel Lock" and many other changes.

For monitoring hotplugging, yes it works. I did it several times today. I have 3 laptops with 3 different GPUs and it works on all of them.

And security?

http://seclists.org/oss-sec/2017/q1/471

It's funny to see a downvote brigade on anything calling into question Linux's supremacy. The overwhelmingly devops HN crowd is probably the worst group to poll on systems software advancement.

Why do downvote brigades even EXIST on this website? I feel like this upvote downvote garbage is an echo chamber in the making. Just sort comments by replies or randomly, or not at all. Or just have a "garbage post" flag and call it a day. Anything is better than this pseudo-democracy thing that Reddit has shown to be useless for real discussion time and again.

Agreed (and upvoted). I think HN has a generally high quality of discussions (except on certain political or fanboyism-related topics where only one side of the debate is tolerated). But, the best discussions on sites I frequent are on blogs that simply sort comments by time and thread, and where the host encourages open and intelligent discussion. Scott Alexander's (https://slatestarcodex.com) is a good example.

The flipside is that "report" buttons can become surrogate downvote buttons, and the lopsided volume of reports leads to inconsistent enforcement of rules, so commenters unpopular with the crowd or mods are more and more likely to be banned, while discussion devolves to groupthink, cliquishness and sycophancy towards the locally-powerful. IME BBSes are the most prone to fall victim to this; the communities are small enough that you can't be "anonymous," and with threads that can continue and be bumped to the top indefinitely, there's no mechanism besides moderator intervention to stop a flame war. Blogs and HN/Reddit-style aggregators do better because old threads are quickly buried.

The success rate for new ground-up modern OS projects achieving broad adoption isn't great - Microsoft's Midori, GNU Hurd, Apple's "Pink", etc. AFAIK there basically hasn't been a broadly successful one since Windows NT in 1993. Which isn't to say people should stop trying and it'll be fantastic if they do succeed with this, I just wouldn't count on it.

It may not be so bad because most of Android apps are built on Java which isolates a lot of what OS does. For other native apps, if Google builds a POSIX compliant OS, it should be fairly easy to port/run.

That's because those weren't completed in time due to feature creep and second system effects (Apple's Pink/Copland), lack of interest and slow progress (Hurd), etc. Midori is just a research project, so noone pushed for it to be adopted anyway.

With a completed OS (which nowadays isn't so difficult, large parts of the stack are either ready or there for the taking), and a big vendor behind it, it's another story. Microsoft pulled it off with NT, and Apple's OS X was a whole new OS as far as Mac customers were concerned (the fact that it already existed as NeXT didn't matter much -- it still required all new software (it didn't run old NeXT stuff as is), new drivers, had a new SDK that Mac developers had to cater too, etc.

Besides Android is not like a full desktop OS Linux -- it abstracts most things for apps through its SDK/"Java", and Linux is just the kernel.

I recall reading a book about the development of NT. One of the key figures joined in because it was seen as the last great chance to create an OS from scratch. Seemed very short sighted to me at the time.

And then Fuchsia will replace Desktops. If the vast majority of devices runs Fuchsia and it comes embedded on Laptops everywhere it will be the end of Windows on the desktop.

I'm OK with that.

Reply on news.ycombinator.com