[HN Gopher] Hardware-accelerated Linux virtual machines on jailb...
___________________________________________________________________
Hardware-accelerated Linux virtual machines on jailbroken iPhone 12
/ iOS 14.1
Author : transpute
Score : 231 points
Date : 2022-06-06 07:55 UTC (15 hours ago)
(HTM) web link (worthdoingbadly.com)
(TXT) w3m dump (worthdoingbadly.com)
| Maursault wrote:
| Since iPhone 4 I have been saying Apple should take over the
| office by replacing all but power users' desktops with an iPhone
| that when at the desk, runs a Microsoft Office/Exchange
| compatible environment (web, email, word processing,
| presentation, spreadsheet, etc.) connected to a bluetooth kb and
| wifi display which all autosyncs along with documents and state
| whenever within proximity as well as seamlessly integrating with
| the present office telephony.
|
| Anyway, this is the best thing I have seen in a while, but the
| demo video would have been more impressive with display sharing
| (with the vm fullscreen) and bluetooth kb&m, because it visually
| proves the viable pocket light workstation.
| swinglock wrote:
| My laptop now connects to my monitor with a single USB-C cable,
| providing power, driving the 4K monitor and connecting to its
| built-in USB hub where the other peripherals are connected. It
| doesn't even have to be wireless because you'd want the phone
| connected to power anyway.
|
| I feel like any recent iPhone would be brilliant for this,
| though the RAM is on the short side, though with slightly more
| in recent Pros it could be a Pro only feature.
|
| Shame it seems like such an un-Apple move to not provide this
| value, that it will likely stay that way.
| astrange wrote:
| You can't run a computer on a iPhone. If you could, it wouldn't
| be iPhone shaped. The fact that you think you can comes from
| the incredible amount of work put into making them fast when
| being used as iPhones.
| jonhohle wrote:
| How is a current iPhone significantly different from an iMac
| architecturally? Sure they are curreny running different
| variants of Apple's operating system, but they share similar
| processors, similar wireless connectivity, etc.
|
| Edit: a slightly different way to look at it: the iPhone has
| always been a computer. Since iPhone 8 or so, it's also had
| desktop class performance.
| paulmd wrote:
| current iphones actually use the same exact chips as the
| macbook air series - the base-tier M1 (although some
| laptops have the more powerful pro/ultra).
|
| the crossover between "high end phone" and "ultra-portable
| laptop" has been observed for a number of years now, that's
| really the space where chromebooks thrived even with
| inferior ARM reference cores, if you throw an actually-good
| M1 into it then you get an extremely solid ultra-portable.
| And the bigger laptops are just more of everything - more
| cores, bigger GPU, more memory.
| MBCook wrote:
| The phones do NOT use the M1, although the M1 is based on
| the chips in the phone. The newest iPad Pro does run an
| actual M1 though.
|
| Either way I agree a recent iPhone would be plenty for
| standard work (documents, spreadsheets, email, web).
| astrange wrote:
| That's the A7 SoC that had a "desktop class architecture",
| not the phone it was in.
| nix0n wrote:
| I used an old Kyocera Rise [0] for document editing a few
| times and it worked fine, and that was a budget phone ten
| years ago.
|
| Having an actual keyboard and a dedicated task switcher
| button made that phone better for work than the modern
| flagship Android I use now.
|
| I don't know much about iphones, but I've definitely seen
| people do real work on ipads, so I think the same argument
| applies.
|
| [0] https://www.phonearena.com/phones/Kyocera-Rise_id7097
| usrn wrote:
| What? I run desktop Linux on a Pinephone which is
| (unfortunately) iPhone shaped. The shape of the actual
| computer inside these devices is often really just a small
| board ~2/3 the size/shape of your credit card anyway.
| 58028641 wrote:
| Why not? An iPhone 13 has a similar single core score to all
| of the M1 processors and a similar multi core score to a
| MacBook Pro (13-inch Mid 2020) with an Intel Core i7-1068NG7
| @ 2.3 GHz (4 cores). However, iPhones only have 4GB or 6GB of
| RAM which may be limiting.
| astrange wrote:
| Because the SoC is not the only component of a computer and
| neither is one benchmark score. Other components are the
| power system, (lack of) cooling, flash storage...
| yywwbbn wrote:
| So an M1 MacBook air basically?
| astrange wrote:
| "Basically" is doing a lot of work there. Those two
| things don't have the same parts in them, just the same
| kinds of parts.
|
| It's not a categorical point but current iPhones don't
| even have real external display support.
| jonhohle wrote:
| iPhones have had external display support for a _long_
| time. Does it require an adapter? Sure. The USB-C only
| MacBooks and iPads do too, unless you have a USB-C
| display. It's entirely possible to display different
| content on the internal display and external display.
| That's an app specific feature in iOS, but not a hardware
| limitation.
|
| Have you seen a current iMac motherboard? Nearly all of
| the I/O is on a breakout board with a thunderbolt
| controller. Does an iPhone expose PCI-E lanes for
| interconnect? No, of course not, but there were desktops
| long before PCI-E and as others have mentioned, the first
| USB-C MacBook had exactly the same number of external
| ports as the iPhone.
| bane wrote:
| Samsung has sorta been trying this for a while with DeX.
| sidpatil wrote:
| And before DeX, there was the Motorola Atrix
| (https://en.wikipedia.org/wiki/Motorola_Atrix_4G#Webtop).
| Un1corn wrote:
| Samsung DeX feels like a POC, sure it works and I've used it
| for about a week while my laptop was being repaired but it
| had lots of paper cuts, most applications were out of place,
| even some of their own.
| ianai wrote:
| Is DeX still being maintained? Like is it available on
| their new flagships too?
| wollsmoth wrote:
| Yup, tried it on a current samsung recently. It's basic
| but it works.
| warning26 wrote:
| The problem here is there's not a clear value proposition. It
| seems like a really cool concept, but who is it for?
|
| What situations exist where a potential user would have access
| to desktop peripherals, but also don't have access to a laptop?
| oarsinsync wrote:
| I've just given up on my M1 Pro MBP in favour of my M1 iPad
| Pro with detachable keyboard case.
|
| I'm very excited to see if the rumours about a proper DE for
| iPadOS when attached to a monitor comes true.
|
| I don't want to carry my laptop around. It's too big and
| heavy. I want to carry one device that I can use as a
| portable device, but also use as a full fat device when I
| have appropriate peripherals, without having to worry about
| some kind of Cloud sync instead.
|
| I don't want to have to wait for state to sync between
| multiple devices. I want a single device that can do
| literally everything. The M1 iPad Pro with detachable
| keyboard (Logitech Folio) and Pencil is dangerously close.
|
| Now if it could run a Linux environment that wasn't
| horrifically slow due to Apple's obscene limitations (see
| iSH, which is incredible, but unfortunately, slow), that
| would solve everything.
|
| For now, I depend on Cloud VMs instead, which falls over once
| my connectivity becomes poor.
| compsciphd wrote:
| part of my phd dissertation was process migration for the
| desktop. I argued that MSFT or apple should make low powered
| portable devices (say phones/tablets) that were your computer,
| but could also be docked to more powerful computers. When
| docked, processes migrate off and run on the powerful hardware,
| when you want to undock, they migrate back.
|
| when docked, the running gui apps switch to a keyboard/mouse
| optimized mode, and when undocked, they switched to a touch
| optimized mode.
|
| I thought microsoft might have been heading in this direction
| with Windows 8 and the RT style apps, but instead of seeing it
| through they got gun shy and went back to more traditional
| style.
| RulerOf wrote:
| I did this back in 2013 I think. I had a desktop and a laptop
| that both supported IOMMU device passthrough. I used the Xen
| hypervisor at the time, and had a VM that I could migrate and
| then hotplug all the PCI devices to between the systems.
|
| I never got it past the "playing around with it stage," but I
| was just _so_ sure that the market was going to move that
| way... Imagine your phone and laptop just migrates out to the
| cloud every time you lock it, then live-migrates back in when
| you wake it up... Or at least between your tiny laptop and
| your beefy desktop. Because nobody would have just one
| computer!
|
| It sure sounds cool to me, but the reality is that it's too
| clunky. Virtualization and live migration are absurd
| solutions to consumer-level problems.
| e-_pusher wrote:
| Surface Book was a step in this direction. The base of the
| computer had a discrete MVidia GPU, so you could run heavy
| GPU loads from the tablet section of the computer while
| connected to the base. If the base was removed, only the
| onboard Intel GPU was available.
| RamRodification wrote:
| > wifi display
|
| Can this ever be performant enough? I feel like even the
| tiniest amount of lag would be unbearable.
| ajconway wrote:
| Of course, people stream VR games over wifi:
| https://www.theverge.com/22218827/oculus-quest-2-wireless-
| pc...
| wlesieutre wrote:
| But does that work in a crowded office building with
| everyone doing it?
| lynndotpy wrote:
| A direct connection with 5GHz WiFi would be ideal and
| realistic. (And, coming up, WiFi 6E and the 6GHz band.)
|
| Contention on 2.4GHz is the main concern I think
| smoldesu wrote:
| Even on 5GHz, your office router would quickly run out of
| bandwidth. A compressed 1080p stream is like 40mbps, so
| even if nobody in the office is using a HiDPI display,
| you'd connect maybe 5 or 6 displays before hitting some
| serious local congestion. You might be able to resolve it
| over ethernet, but if 'wireless' is the name of the game,
| then I'd suggest looking elsewhere.
| ac29 wrote:
| A high quality 1080P video stream is more like 5-10Mbps,
| and a typical desktop use case is going to be much, much
| more compressible.
| wlesieutre wrote:
| They did say "direct" so the router wouldn't be a concern
|
| https://www.wi-fi.org/discover-wi-fi/wi-fi-direct
| swinglock wrote:
| "Direct" Wi-Fi still shares the same spectrum though, so
| even without central points it's not equivalent to
| cables.
| wlesieutre wrote:
| Right, that's the part of it I'm worried about in a
| crowded building
| bee_rider wrote:
| But it is a fundamentally different type of problem.
| Going through a router imposes hard bandwidth limitations
| at a central point.
|
| With direct wifi, each monitor only needs to serve a
| single user, so the devices can be lower powered -- the
| broadcast only needs to serve ~a single desk. And they
| are centrally controlled by the IT department, so unlike
| a wifi router the in RF mad-max anarchy zone that is a
| modern apartment building, there shouldn't be concerns
| about noisy neighbors.
|
| I'm not sure there's a point to all this -- the phone
| being used as a computer is probably going to be plugged
| in, so why not push everything through USB-C/thunderbolt
| nowadays? -- but it is almost certainly possible.
| swinglock wrote:
| The difference between sharing limited spectrum and AP
| resources in a room and only sharing limited spectrum
| resources in a room are not a fundamentally different
| problems. In addition you'd have loads of noisy
| neighbors, at least everyone close by in the room.
|
| I agree the phone ought to be plugged in anyway. Even if
| it could survive a full day of being your workstation,
| you wouldn't want a half dead or worse battery at the end
| of your work day, and regularly draining a significant
| amount while stationary unnecessarily shortens the total
| battery lifespan. In addition, if docked into something
| with a bit of active cooling, the phone could perform
| better and once again preserve the battery.
| bee_rider wrote:
| I think they are fundamentally different. For the first
| case, everyone _must_ broadcast with enough power to
| reach the AP. It introduces a bottleneck.
|
| For the second case, everyone only needs to broadcast
| with enough power to reach their device. Overlapping
| coverage areas are essentially a side effect.
| smoldesu wrote:
| I've used 3 different wireless VR implementations before,
| all of them require absurd amounts of bandwidth
| (50-130mbps), and send a highly-compressed image. It looks
| worse depending on the display you're using; on the OLED
| display of the Quest, the image ended up looking
| particularly terrible.
|
| Even with proper display stream compression, I doubt you
| could build a network robust enough to host more than 2 or
| 3 wireless monitors on one network.
| bee_rider wrote:
| VR required like 90fps at something >1080p, right?
|
| Office work is fine with 30fps (can realistically drop
| lower). Also I suspect office workloads should be
| compressed more easily (think of a next editor -- lots of
| white-space, only a couple characters changing from frame
| to frame).
|
| And peer-to-peer rather than through the router would
| probably be the way to go. Since 5GHz drops off pretty
| quickly, it might be possible to exploit this if people
| aren't packed in too tightly. Which gives a nice, hard
| technological backing to requests _not_ to be crammed in
| too tightly.
| jsight wrote:
| Its probably quick enough now, but tbh, I'd rather it just
| run over USB-C. Maybe then it could keep the phone charged up
| at the same time too.
| hwhwhqoabsg wrote:
| Can someone please explain to me why a jailbreak is needed for
| running a VM?
|
| I'm just why it wouldn't be possible to run a virtualized OS as a
| native app, and translate all OS calls into calls that the native
| app could make to the underlying iOS.
|
| I'm wondering what the restrictions are?
| meibo wrote:
| It's not needed, but as you can read in the article, Fedora
| takes half an hour to boot without virtualization or JIT. You
| cannot mark memory executable in iOS apps, preventing you from
| actually letting the host CPU execute any guest code.
| sgjohnson wrote:
| UTM can actually be installed using AltStore (which doesn't
| require a jailbreak!)
| Nextgrid wrote:
| The key here is the hardware acceleration - the default UTM you
| can install out of the box doesn't do that.
| easton wrote:
| But I wonder if you could use the modified version the author
| posted with AltStore (or plain old Xcode signing) to get
| virtualization support on the M1 iPads which don't require
| kernel modifications. I'd pay $100 a year for a developer
| account if I knew I could run macOS and Linux at near native
| speed on my iPad.
|
| Does anyone know if you can install apps that use private
| entitlements without a jailbreak if you compile/sign them
| yourself?
| NathHorrigan wrote:
| I assume the performance improvements come from having JIT
| memory enabled? If so this needs to be done by starting the
| app in 'Debug' mode from XCode which loads a specific
| entitlement.
| GekkePrutser wrote:
| Sadly this will never fly without jailbreak due to apple's ban on
| emulators and the like.
| londons_explore wrote:
| If this gets popular enough and lots of users start demanding
| it, then apple will edit their policy.
| throw457 wrote:
| This is such a small edge case it would never profitable to
| implement.
| usrn wrote:
| Lol. I remember still having faith in Apple.
| madeofpalk wrote:
| I mean it'll never fly due to apple's ban on _apps modifying
| the kernel_.
| webmobdev wrote:
| A better use of this would be to run one of the Linux mobile OSes
| - for example, Sailfish OS which has tablet support too (would be
| great on an iPad).
| rcarmo wrote:
| The most interesting bit for me is that iPads currently support
| virtualization, both in the chipset and the kernel.
|
| All that wasted compute power could be used to run a developer
| sandbox if Apple wanted to (iSH is great, but it's sandboxed x86
| emulation and not quite all there yet for back-end development).
| nicoburns wrote:
| It could presumably also be used to run a completely sandboxed
| copy of macOS on the phone while simultaneously running iOS
| (for use when docked with a mouse/keyboard/screen).
| 2OEH8eoCRo0 wrote:
| This was my vision- macOS as an app on M1 iPads. I think
| they're obviously afraid to give users a way around the app
| store.
| Gigachad wrote:
| I think it's more likely that they want products to create
| touch native versions of apps while allowing running macOS
| apps directly would result in less iPad apps being made
| since telling the user to connect a mouse and keyboard is
| workable.
| pornel wrote:
| OTOH they've allowed iOS App Store apps to run on M1
| macOS -- despite MacBooks not having a touchscreen.
| layer8 wrote:
| Operating iOS apps with a mouse/trackpad is not a
| problem, but operating macOS with a touch screen is.
| londons_explore wrote:
| It tells me that Apple is big enough to have a team of people
| who developed a feature which seems to work fine (hardware
| virtualization) but was never launched.
| mschuster91 wrote:
| I'd guess that it made sense to use iPads as a cheap testing
| ground for stuff before they went all-in on M1. Maybe they
| found bugs in hardware that they could fix in time for the M1
| release but made virtualization impossible or a security hole
| on iPads.
| Nextgrid wrote:
| M1 MacBooks are basically iPhones with a big screen and a
| keyboard. I doubt they explicitly tested or planned for this
| to work on iOS.
| meibo wrote:
| The post does state that it's explicitly disabled for
| iPhone builds of iOS, while it's basically enabled for iPad
| builds - you just need the entitlements.
|
| Maybe some Apple engineers are running around with an
| internal VM app running Fedora because they were tired of
| only using their powerhouse iPads for Netflix and
| scribbles?
| mhh__ wrote:
| Now you can use your M1 to remote into a single core 2GB cloud
| box 5000 miles away!
|
| Progress!
| frou_dh wrote:
| As seen on HN in the half a dozen "I'm a software developer
| and now I do all my work on my iPad*" blog posts over the
| years.
| xattt wrote:
| This could have been Plan B for Apple if governments forced
| their hand in side-loading apps.
|
| This makes me think of the way Intel shoehorned the x84-64 into
| Prescott Pentium 4 chips, enabled with a BIOS update when
| Athlon 64 released back in 2005.
| madeofpalk wrote:
| "All that wasted compute power" has been the story of the iPad
| for the last 5 years. Very powerful hardware that simply isnt
| used by the software on it.
| Infernal wrote:
| Trying to look at the other side, this is how you still
| support machines like the iPad Air 2 released in 2014...
| madeofpalk wrote:
| iPad Air 2 review in 2014
|
| > _If the hardware of the iPad Air 2 demonstrates the
| overwhelming power of small iterative improvements, then
| the software represents the failings of that approach. The
| overall experience of using the iPad Air 2 in 2014 is a
| case study in missed opportunities and untapped potential.
| Apple has all but stopped adding tablet-specific features
| to iOS -- the minor two-paned mode for landscape apps on
| the iPhone 6 Plus is a more significant rethinking of how
| to manage a larger screen size than anything added to the
| iPad Air 2 this year._
|
| > _Just consider something as simple as browsing the web.
| On raw benchmarks the iPad Air 2 is comparable to a 2011
| MacBook Air -- which, again, is crazy -- but the MacBook's
| version of Safari is vastly more feature-rich and flexible.
| That MacBook will also allow me to run multiple apps
| alongside Safari and be far more productive than the iPad;
| we're well past the point when Apple needs to figure out
| proper multitasking on its tablet._
|
| https://www.theverge.com/2014/10/21/7027485/apple-ipad-
| air-2...
|
| _The_ story of the iPad is that Apple is shipping hardware
| vastly more capable than it 's software.
| rhn_mk1 wrote:
| I'm not sure if that other side is positive if it means
| that we need more power to give user the same functionality
| as before.
|
| I imagine that the basic uses haven't really changed since
| 2014: write to others, video calls, read stuff on the web,
| watch videos, use web apps where expensive logic is
| outsourced to a remote server.
| jsight wrote:
| I was just thinking of this yesterday. The ipad could make a
| great dev machine if I could run a rooted Linux environment
| on it.
|
| Although a little more RAM would be nice too.
| aiddun wrote:
| I'd love to be able to use an M1 iPad Pro as a remote
| compilation server
| rollcat wrote:
| I think it might be coming, but of course on Apple's terms.
| Swift Playgrounds has been a thing for a while, now you can
| make native iPad apps on an iPad. It's not going to be the
| hacker's wet dream, but I wouldn't be surprised if a future
| update makes use of the hardware.
| moonchrome wrote:
| This is exactly why I'm hesitant about going back to Apple
| ecosystem. I like the M1 stuff and the recent hardware
| changes - but the walled garden and the Appstore lockdown to
| protect their cash cow are coming at the expense of
| functionality and I just can't get behind that.
|
| Hoping the next gen of x86 chips finally hit that sweet spot
| of performance/low power/heat.
| prox wrote:
| The cash cow is also what keeps the quality and user
| experience high.
| peoplefromibiza wrote:
| _but the walled garden and the Appstore lockdown [...]
| are coming at the expense of functionality_
|
| I guess this comment says that the user experience is
| going down, _because of_ the cash cow and despite it 's
| generating a huge amount of money.
| prox wrote:
| Not my experience. Maybe for specific stuff for HN type
| users, but in general the functionality (aka options) in
| recent years have gone up, not down. Most apps are very
| high quality compared to Android especially (and yes I
| use both ecosystems)
|
| There is a reason it _is_ a cash cow. People like buying
| it.
| peoplefromibiza wrote:
| Sorry if this sound dismissal, but Ferraris are a cash
| cow too.
|
| People like to buy them.
|
| _Research shows that there are 80+ apps installed on the
| average smartphone. But with that said, people aren 't
| using all of those apps. The average person uses 9 mobile
| apps per day and 30 apps per month._
|
| Most of them are games, the apps people can't live
| without are ~15 and are made by largest brands of social
| media platforms and e-commerce
|
| They have the resources to make the same high quality
| apps on every platform.
|
| In my view Apple users will keep using Apple by the force
| of habit, Apple users will perceive lower quality in
| Android ecosystem, because there's more choice.
|
| Also, on Android there are a lot of open source apps that
| are uglier but work exactly as intended, no ads, no phone
| home etc
|
| If you go to a Bang & Olufsen shop and then to RadioShack
| the perception is of a lower quality, which is probably
| true, but who needs bang & olufsen headphones to listen
| to a 64kbs podcast while driving to work?
| FpUser wrote:
| My "user experience" includes using hardware / software
| to it's potential extent. If I am limited to a
| kindergarten state for marketing reasons I simply do not
| join the ecosystem.
| usrn wrote:
| No. The user experience on iOS when you want to use any
| kind of FOSS (xmpp and irc clients, compilers, git
| without some closed wrapper etc) is _shit._ You 're often
| stuck without notifications and you might even have to
| use iSH which is ~1/10th-1/100th speed of the original
| Pinephone.
|
| It also _doesn 't stop malware_ since the contractors
| doing app reviews don't instrument the apps, they just
| play with them for a few minutes to see if something
| looks wrong. The App Store is full of malware but it
| makes you feel safe.
| HidyBush wrote:
| FOSS on Apple products that only have an app store is
| doomed because it costs money to have a developer
| account.
|
| FOSS activists always claim that it's free as in free
| speech and not free beer, but somehow these FOSS apps
| everyone claims to need can't even reach the $100/y
| necessary to keep them on the app store.
| usrn wrote:
| Not just the developer account, the developers are
| _required to maintain app specific infrastructure_ to do
| things like push notifications. That 's why FOSS apps
| rarely have them.
|
| All of this on top of the the hardware cost and having
| someone run the iOS API treadmill and deal with app store
| reviews makes FOSS on iOS prohibitively expensive for
| everyone but the most committed. It pretty much filters
| out everyone that isn't a corporation or irrationally
| passionate.
| HidyBush wrote:
| I mean, I would suppose people who want IRC on their
| phone must be pretty senior. They can't spare 2 dollars
| per year to pay for the developers to keep the app
| afloat?
| zaik wrote:
| Monal seems to be a kind of okay XMPP client for iOS.
| usrn wrote:
| Exactly. It's all "kind of okay" at best. OSX used to
| have the best XMPP and IRC clients and some were ported
| to iOS; it's not a lack of effort on the part of the
| developers it's a critical problem with the platform.
| rollcat wrote:
| > Hoping the next gen of x86 chips finally hit that sweet
| spot of performance/low power/heat.
|
| Well I'm hoping we're finally getting the shove to move off
| x86, there's nothing particularly exciting about the
| architecture itself, and we're finally getting a tangible
| proof that the compatibility layer has a huge price tag.
| I'm excited because displacing x86 is also a stepping stone
| for RISC-V, and might push various vendors to consider
| making their software more portable - which would be a win
| for literally everyone.
|
| > [...] the walled garden and the Appstore lockdown to
| protect their cash cow are coming _at the expense of
| functionality_ [...]
|
| Emphasis mine - what is the functionality trade-off that
| you're referring to? The App Store has never stopped me
| from running any app from outside the App Store.
|
| I would agree that macOS is continuously getting a bit less
| hackable over time (how do you authorise cron to access the
| downloads folder...), but this has very clearly everything
| to do with the half-assed attempts at making the system
| more secure (a very noble and respectable goal), and
| nothing to do with restricting user freedom (after you
| factor in incompetence, carelessness, and ignorance).
|
| OpenBSD has always been making broadly similar moves:
| securelevel, signing packages, pledge/unveil, removing
| support for loadable kernel modules, removing
| unmaintained/insecure subsystems (Linux emulation,
| Bluetooth). Some of these have been a bit annoying, but
| reading this as removing user freedom seems like mis-
| interpreting the intent.
| smoldesu wrote:
| > Well I'm hoping we're finally getting the shove to move
| off x86, there's nothing particularly exciting about the
| architecture itself...
|
| I actually largely agree with you here. We need to be
| transitioning to open ISAs, so getting off x86 isn't
| something I disagree with. _However_ , jumping from one
| proprietary ISA to another, slower one doesn't exactly
| make sense to me. No matter how you frame it, x86 is
| still the performance and compatibility king. I'd love a
| RISC-V laptop as much as the next guy, I think this is
| more about Apple increasing their profit margins by re-
| using their ISA licensing from their phones/tablets.
|
| > Emphasis mine - what is the functionality trade-off
| that you're referring to? The App Store has never stopped
| me from running any app from outside the App Store.
|
| Well, the discussion is the Apple _ecosystem_ , not just
| Mac, and in that sense the App Store does prevent you
| from ingesting third-party software or running arbitrary
| code on your own device.
|
| I also don't think this is about security as much as it's
| about control. Apple now directly competes with companies
| like Spotify, Netflix, Hulu, Microsoft, Meta, the list
| goes on for days. If they control their hardware and
| software experience end-to-end, then they control the
| users too.
|
| > ...and nothing to do with restricting user freedom
|
| But how exactly can we prove that? Apple says one thing
| and does another. They say they're protecting user
| privacy by hashing your cloud photos with a neural
| network. They say they've built the fastest GPU and then
| backpedal when it gets benchmarked. If they give you less
| capabilities with each update and say it's making you
| safer, I don't think I'd take their word _entirely_ at
| face value; maybe security is _part_ of the motivation,
| but almost certainly not the entire story.
|
| As for OpenBSD? I actually like their approach for the
| most part, but there's a reason it's still a fairly niche
| OS. People like having options, and taking away your
| Linuxulator and Bluetooth modules doesn't make for a
| super enticing desktop experience. MacOS suffers the same
| issue, but more for the development crowd. There are so
| many footguns in MacOS for developing standard *NIX
| software that it makes my head spin. Want the standard
| GNU coreutils? Of course they're not built in, go grab a
| package manager and download them posthaste! Want to use
| Git? Here, download 750mb of random utilities onto your
| system so the 20mb program can execute. Oh, you wanted
| Bash to be up-to-date? We sorta forgot about that one...
|
| ...but their MacOS, iPad and iPhone development tools get
| fixed just fine. Gotta tend to the breadwinners before
| you take care of the intrinsic issues in your OS, I
| guess.
|
| In closing, I don't actually disagree with a lot of your
| rhetoric, but I don't think you're addressing the
| parent's statement. Apple is flighty, they do crazy
| things in the blink of an eye and everyone else is forced
| to just go along. What happens if I go back to the Apple
| ecosystem and they remove all the iPhone and Macbook
| ports? There could be so many "gotchas" that many people
| simply don't feel safe going back to that lifestyle
| anymore. There is no roadmap, and that scares people who
| rely on stable software for a living. It's why I end up
| developing more on Linux than I do on Mac.
| rollcat wrote:
| > No matter how you frame it, x86 is still the
| performance and compatibility king.
|
| I'm not up to date with the latest benchmarks, but last
| time I checked, the M1 family is near the top in raw
| performance, and completely destroying everything else in
| perf per watt (CPU _and_ GPU) - at competitive prices.
|
| The compatibility angle is pretty interesting. I've been
| using an M1 Mac for about a year, and the only time I
| notice that I'm not on an x86 system, is when I run
| Electron apps compiled for x86 only. (I think it shows
| just how bad of an idea it is to ship JIT kitchen sinks
| as "native" apps, but it's an entirely different topic.)
| Surprisingly enough, the ARM build of Windows 11 in
| Parallels _also_ does a pretty good job of running x86
| software, including even some pretty dated games (like
| Soldat). GPU works really well too - Edge in Parallels
| beats native Safari on shadertoy.com.
|
| Of course I'd rather see software that is actually
| portable and built natively, but honestly, I'm just
| impressed with what these companies have pulled.
|
| > I think this is more about Apple increasing their
| profit margins by re-using their ISA licensing from their
| phones/tablets.
|
| I don't mind Apple increasing their profit margins, if
| other parties benefit as well. Consider the impact on the
| wider software ecosystem. Software is usually difficult
| to port to a second platform, because the developer was
| hardcoding some assumptions. Often porting to a secondary
| architecture, OS, or platform will lay the groundwork to
| remove many of these assumptions, and enable porting to
| other systems. Portability tends to contribute to
| software quality, and having more software available
| benefits the users of the secondary platforms. As I said:
| everyone wins.
|
| > [...] in that sense the App Store does prevent you from
| ingesting third-party software or running arbitrary code
| on your own device.
|
| This is technically incorrect. You don't need the App
| Store to run arbitrary code on your iPad/iPhone - you
| only need a Mac and XCode. Which is either better or
| worse, depending on your POV.
|
| And again - you can now write iPad apps on an iPad. You
| still need to jump all the hoops to distribute it on the
| App Store, but if you consider the work it takes to e.g.
| have it included in Debian repositories, or OpenBSD
| ports, it's roughly equivalent. "You can bring your toys
| to my playground, but we still play by my rules."
|
| > They say they're protecting user privacy by hashing
| your cloud photos with a neural network.
|
| I 100% agree with you on this one. The premise looks
| innocent: iCloud Photos is opt-in, and Apple is doing
| Apple by moving their CPU-bound workloads away from
| clouds and onto user devices. Technically this is fine:
| nothing is being scanned that wasn't already being
| scanned server-side. But just having the mechanism at all
| is scary, it has a lot of potential to bring harm, and
| it's a very dangerous precedent.
|
| > If they give you less capabilities with each update and
| say it's making you safer, I don't think I'd take their
| word entirely at face value [...]
|
| I'm not taking that at face value, but I will blame their
| "honest" incompetence, carelessness, and ignorance before
| I will accuse them of malice. They often rewrite stuff
| for the sake of rewriting it, and break "edge" cases that
| thousands of people rely on. The cries the the thousands
| drown in a billion sales though.
|
| https://duckduckgo.com/?q=site%3Ajwz.org+fucking+apple
|
| > Want the standard GNU coreutils? Of course they're not
| built in, go grab a package manager and download them
| posthaste!
|
| Same on every _BSD. macOS is just staying true to its
| roots ;)
|
| > Oh, you wanted Bash to be up-to-date?
|
| Notice that macOS' Bash is the final GPL-2 release. This
| is entirely on FSF working overtime to make
| redistributing their software more annoying. One of the
| reasons why I'm a big fan of the _BSDs, is that they take
| this no-bullshit approach to redistribution: the license
| has to answer one simple question, "can I use this?",
| preferably in a few very short sentences.
|
| Personally, I'd stay away from Bash. If an otherwise
| POSIXly kosher shell script becomes hairy enough to call
| for Bash's extensions, I will label it a monster and
| rewrite it in Python. ZSH and (OpenBSD) KSH both make
| much nicer interactive shells too.
|
| > Apple is flighty, they do crazy things in the blink of
| an eye and everyone else is forced to just go along.
|
| Agree on that. "Our playground, our rules", often taken
| to an extreme. Which is also why I like voting with my
| wallet, and keeping an alternative around (OpenBSD on my
| laptop). If I don't like an Apple toy, I simply don't buy
| it.
|
| > It's why I end up developing more on Linux than I do on
| Mac.
|
| Which explains why you're having such a hard time with
| systems that don't include the non-standard GNU or Linux
| extensions. I develop and/or test across five different
| OS+arch combinations, which often catches bugs and
| hardcoded assumptions. I often wonder if it's worth it,
| but then remember how monocultures tend to just rot.
| smoldesu wrote:
| > the only time I notice that I'm not on an x86 system,
| is when I run Electron apps compiled for x86 only.
|
| Maybe it's just from working in devops at a Mac shop, but
| getting dev environments working on M1 with deployment
| parity is nearly impossible. It requires vast redesigns
| of preexisting architecture and forces a lot of software
| off of the table. It's wholly a YMMV situation, but I've
| seen things on both side of the fence; I still reach for
| x86 when I'm getting work done.
|
| > Portability tends to contribute to software quality
| [...] everyone wins.
|
| I don't really mind that aspect. What I _don 't_
| appreciate is Apple raising the walls around their
| garden; Sure, you can claim it's being done to make
| people safer, but as someone without much skin in the
| game, this just makes Apple look worse to me.
|
| > You don't need the App Store to run arbitrary code on
| your iPad/iPhone - you only need a Mac and XCode.
|
| I mean, at that point it's no better than a
| microcontroller.
|
| "Want to run your code? No problem, all you have to do is
| plug it into the ROM programmer!"
|
| My gripe is that Apple goes out of their way to prevent
| people from distributing software anywhere that they
| might not profit from/control. If I buy a computer, I
| ought to be able to run the software _I_ want. That logic
| extends to phones and every other device in their lineup,
| for that matter.
|
| > But just having the mechanism at all is scary, it has a
| lot of potential to bring harm, and it's a very dangerous
| precedent.
|
| Sure, I don't think many people will disagree here. The
| posturing here is what scares me, and it's the sort of
| sociopathic "we know what's right for you" behavior that
| turns me right off.
|
| > I will blame their "honest" incompetence, carelessness,
| and ignorance before I will accuse them of malice.
|
| I would too. One problem though: they seem to have a hard
| time admitting when they're wrong, and insist that
| everything they do is deliberate. Remember "you're
| holding it wrong"?
|
| > Same on every BSD. macOS is just staying true to its
| roots ;)
|
| I would have preferred an "it just works" solution, but
| I'm sure that's the answer most Mac users would issue me
| so I'll take it.
|
| > Personally, I'd stay away from Bash. If an otherwise
| POSIXly kosher shell script becomes hairy enough to call
| for Bash's extensions
|
| I do, I use Fish as an interactive shell. But running
| recent bash is a requirement for maintaining servers,
| cross-platform *NIX development and a huge number of
| toolchains. I don't like it, but the ease-of-installation
| partially predicates how difficult it is to set up my dev
| environment.
|
| > If I don't like an Apple toy, I simply don't buy it.
|
| I mean, I just take this ideology to it's logical
| conclusion; if I presently think Apple is being scummy, I
| don't even give them my money. Explains why I haven't
| paid for an app since 2013...
|
| > I develop and/or test across five different OS+arch
| combinations, which often catches bugs and hardcoded
| assumptions. I often wonder if it's worth it, but then
| remember how monocultures tend to just rot.
|
| I write/test across multiple arches as well, I just end
| up writing the code on Linux. It's really not worth it to
| use Mac when you're only actually deploying to
| x86_64-Linux.
| zekrioca wrote:
| Not related to the video, and I know this may read as a not so
| polite thing to write. However, the person in the video should
| probably clip their nails. And it is not just about the
| appearance, but also about their health risks [1].
|
| [1]
| https://www.cdc.gov/healthywater/hygiene/hand/nail_hygiene.h...
| fartcannon wrote:
| Think of the power that is thrown away yearly by Apple. All these
| locked down computers capable of so much being used to run what,
| a camera app? Lifetimes of wasted potential. It's a crime against
| the future, if you ask me.
| usrn wrote:
| The real crime is the damage to the next generation's minds.
| Their expectation that all meaningful software must have an
| organization and online infrastructure behind it and the
| general infantilization of computer users is honestly kind of
| terrifying.
| teaearlgraycold wrote:
| At least it's less job competition for me.
___________________________________________________________________
(page generated 2022-06-06 23:01 UTC)