Hacker News new | comments | show | ask | jobs | submit login
Leaving htop running freezes macOS High Sierra (github.com)
332 points by alpb 10 months ago | hide | past | web | favorite | 233 comments



High Sierra seems like a great OS if you want an unstable workstation that doesn't work with tools you've used everyday for well over a decade.


I came for the filesystem bugs, I stayed for the absurd root exploit, but its the little crashes, rendering glitches, and feeling of general instability that made me fall in love with High Sierra...


APFS has been nothing but pain for me. File system changes show up in Finder minutes later! I should have stayed in Sierra.


Unfortunately, it seems Apple has forced us to choose between Meltdown mitigation and a somewhat working OS. The recent Sierra security updates didn’t include the Meltdown patch.

https://twitter.com/theregister/status/949358083431546880


Apple has actually backported the fix to Sierra and even El Capitan. https://support.apple.com/en-us/HT208331


The tweet in the post you replied to points out that the page that you linked to was altered today to no longer say the fixes extended to those two.


My educated guess, based on 14-years working for MIT’s IT department: the fixes for Sierra and El Capitan broke something important, so they got pulled.

When they get fixed, they’ll be re-released…


My MacMini6,2 (2012 Server) has two SSDs RAIDed together. High Sierra beta worked fine and upgraded fine to golden master and subsequent. When I decided to reinstall my OS from scratch and reload from backups I discovered to my horror that mine was not a supported configuration and there was no way in hell I would be getting back to where I started from. This happened on the 20th November. I finally managed to boot my system again on the 20th December. After failing to re-enable RAID I broke my system into individual root and /User partitions on separate drives (or tried to) and discovered that APFS (Encrypted) and /etc/fstab don’t get along. I re-defined my home folder to be on the second drive but cannot log in until I’ve logged into a utility account and mounted the encrypted partition separately. Currently permissions are so fucked up I can’t even stream music from iTunes. On Monday I’m gonna nuke it again and try another route from scratch.

Throughout this Apple Support has been anything but (aborting when for some reason my Mini’s S/N failed to match their records for Server machines, and thus bailing on me on the premise that I shouldn’t have RAID anyway so basically screw me.)


I've felt for a long time that Apple have lost interest in their end-user computers. MacOS Server has long seemed like 'if you REALLY want it, here, have fun'. But the tools tended to work. SoftRAID somehow works fine on Windows, why not on Mac.

Can't imagine how much pain you've been through with this. Got enough backups to revert back to Sierra?


And to think that ZFS for FreeBSD had already been up and running for years...


Not on 1B devices.


Maybe not on 1 billion devices, but ZFS was running on mission-critical Sun/Oracle hardware with expensive support contracts for many years. ZFS is currently probably the most reliable modern filesystem.

Still, rolling out APFS on such a large number of iDevices/Macs is a very impressive feat (although APFS is, of course, a simpler filesystem). The problems are with many of the 'edge cases' (RAID, etc.). This is not really surprising, btrfs has been in development since 2007 and still has problems with more complex setups.


Secret hint: if it's an older mac, try the NVIDIA Web Driver for the video driver.


It's a 2017 MBP with touchbar (and a Radeon GPU).


Rendering glitches when in Chrome are awful indeed and happen way too often


I'm on macOS High Sierra for the spontaneity. Every morning, I go to wake my Macbook Pro with two 4K USB-C connected displays and external keyboard and mouse. Will one display turn on? Both? Neither? Who knows!


Store the screen layout using serials from the EDIDs. Why it doesn’t do this and “tries” to guess their position somehow is beyond me. I use the OWC TB3 dock with dual Asus PA328qs with the latest firmware. Because of apparently some lame design decisions, in order to run dual external monitors over a single TB3 port, you need the dock, which has one DisplayPort output, then a TB3-> DisplayPort (via USB-C alt mode IIRC) dongle plugged into the dock. This is idiotic, but I wouldn’t mind so much if it actually worked consistently. This is not the case. Half the time, when I go into work and plug in, I get 30hz sync on the monitor plugged into the dock. This is when it works at all, without having to guess if the laptop is actually on, because it’s the rMBP which has no power indicators despite having a useless oled screen across the top of the keyboard. So, plug in, wait 5-30 seconds, nothing, open laptop, nothing, unplug dock, nothing, hold down fingerprint reader (the idiotic power button on the rMBP) until...nothing happens, then wait a random amount of seconds, stab it again, wait another 5-10 seconds for the Apple logo to appear, macOS boots into recovery (reset credentials screen, because it assumed that since I couldn’t log in during a boot that I need to reset my password I guess), restart, wait for disk decryption screen which doesn’t work with my wireless keyboard plugged into the dock, type in that password on the laptop keyboard, wait another few seconds for it to log in, ensure that WiFi is enabled as my normal network connection is via LAN via the dock which isn’t plugged in at this point, log in, wait for everything to start, close lid and pray that I don’t have to repeat this, wait another 5 seconds for it to actually sleep, plug dock in, mash random keys until it wakes up, log in again, rearrange monitor layout as it’s usually swapped L/R, one monitor doesn’t work, or they both don’t work again, fix that by power cycling monitors / dock, then finally maybe do some work if it’s not yet lunch time.

For a ~$6k setup, it’s awful. It’d be awful for $1k. About 10 years ago, I had a Dell D830 running Fedora and later, early versions of Gnome shell. The dock powered the laptop and a mouse, keyboard, external drive, and two monitors. It was a single (albeit proprietary) connector, and the dock was fairly expensive (~$200 IIRC). It worked damn great, nearly every time, even with sleep States in Linux on a laptop 10 years ago.

Why is this suddenly harder than putting a Tesla on Mars? Why don’t we have good real wireless HD displays by now? The current state of affairs with Apple laptops and docking is just sad to me. I’m almost bothered enough by it to switch jobs to go there and work on fixing it.


I hit a breaking point with the Apple laptops a couple of years ago. It was actually iTunes that drove me off. iTunes Match somehow corrupted and lost a bunch of my music. Forever gone and I didn't have backups. After working with Support for a while, I finally gave up and bought a new ThinkPad T470s and put Arch Linux on it and never looked back. Things work really well and predictably. I have it dialed in just how I like it; every hardware component except the proprietary fingerprint reader is supported. Switching to a tiling window manager has been a huge productivity boost, too.

Now that office suites are in the cloud along with music and everything else, I no longer see the draw of Apple hardware for a developer laptop.


Someone did go through the trouble of reverse-engineering the fingerprint sensor. The code works on my thinkpad. https://github.com/nmikhailov/Validity90

Only thing left is someone willing to clean up the code and upstream it.


Was it hard to move off Retina and reasonable touchpad? I am considering similar route, but these are the two things that keep me on Apple.


High DPI displays are common on PC laptops now, they generally work well in Linux (not so much for legacy apps but you're unlikely to use those).

I like the touchpad on my X1 Carbon, but I never used a mac long enough to really adapt my workflow to all the touchpad gestures (I think on Linux you'll be happy with general usage/scrolling/etc, but stuff like pinch to zoom is still not widely supported).


High DPI displays are common on PC laptops now, they generally work well in Linux (not so much for legacy apps but you're unlikely to use those).

Surprisingly, some Tk applications that I use irregularly actually work with HiDPI with Tk 8.6 (as in, they are scaled up with an integer scaling factor, and are sharp).

Multiple displays with different DPIs also work with Wayland (I have one 4k screen and a LoDPI display 90 degrees rotated). Though, there are some bugs with GNOME 2.26 on Wayland with HiDPI displays that are switched off and on again (e.g. some applications such as GNOME Terminal become blurry, they seem to be rendered at lower DPI and scaled up). I have reported some bugs and the gnome-shell/mutter developers are quite responsive.


I got used to the touchpad with time and the touchpad drivers have improved, too. It's really not a big deal at all after a few weeks.

HiDPI is generally pretty decent although it's slightly annoying that different layers of the graphics stack (Xorg, GTK2, GTK3, etc.) have different mechanisms for tweaking the scaling but with a few evenings of work, I have everything dialed. A small price to pay for years of use to come. The Arch Linux wiki is super helpful for this.


reasonable touchpad?

You want to look for windows precision touchpad support. Turns out the problem with windows touchpads was not the hardware, but the crappy drivers touchpad vendors make. Microsoft made their own drivers and if you can get hardware compatible with those you'll have a much nicer experience.


The recent thinkpads have hi-dpi displays, and they're perfectly fine for anything other than intensive colour work.

The touchpads are fine too. I regularly bounce between a macbook and a T460, and I don't have any problems operating the latter.


If you choose a ThinkPad, the trackpoint, three mouse buttons and the perfect keyboard together are such a nice combination I actually disabled the touchpad from bios.


My Yoga 2 Pro has hidpi display and better touchpad than MBP's.

Yoga 900's touchpad is way worse though, so if you really care, always check if you can try it first.


You can get higher resolution displays on other laptops. As for the trackpad most are pretty good now days, although apple is probably still in first place.


I stuck with the hardware (albeit quite old) and went linux for similar reasons.

The older hardware can be quite nice with linux, but my next notebook wont be a mac, not due to value for money (spec and built quality balance is actually quite good for price these days), but because all attempts to run linux or any OSS on the more recent models are a disaster.


Try a modern ThinkPad with Intel graphics. I promise, it will work great.


I know it's a traditionally good brand for compatibility, but for me they sort of dropped quite far from the top of the list since becoming a chinese owned company and for superfish etc. (To clarify my problem with it being chinese is the prevalence of injecting backdoors into absolutely everything from state or otherwise, and these days we have to worry about hardware too)

If I was to buy right now my list is roughly in the order of:

    - Dell XPS 13/15
    - Razor Blade (It's getting official linux support soon)
    - Librem 13/15
    - One of the million and one thinkpad models although I have no idea which
I'm most likely to just wait longer though :P the only thing to die on this old thing so far is the keyboard springs out of sheer wear and the battery. I'm gona replace the keyboard soon myself.


I see the draw for desktop and mobile developers.

Developer is not a synonim for UNIX programming.


Hey I’m not the only one to have Match corrupt my library on apples servers and lose music?!

Lucky for me I have a copy of all of it on my 2011 iMac hiding in my closet.


[flagged]


That's like asking someone with Dropbox storage to make sure they have a secondary backup on an external HDD somewhere. It's not that it wouldn't be a good practice, it's that Dropbox should be the one making the backups of their backups.


And you should be following the 3-2-1 rule for backups.

3 copies of the data, 2 locally on different media, 1 offsite.


They do do that. It fails because the monitors have identical EDIDs. Peripheral vendors are less competent than you think.


Uuuh that shit again. Had to write code long ago that tries to set outputs to clone if one of them is a projector (on X11). Seemed simple at first since it seems to be agreed upon that projectors report a physical size of 0x0 via edid. Which makes kinda sense since you don't have a fixed image size. Until I discovered that some projectors don't because the vendor thought it would be better to report an arbitrary value. Then you got setups where there's a middle box that swallows edid or just reports other random data.


Perhaps someone can make an adapter that can sit between the computer and monitor to fix EDID issues?


It has been done for the T221. It's quite a hack, expensive and error prone.


> Store the screen layout using serials from the EDIDs

What does this mean/how do you do it?


EDID is a little EEPROM inside virtually all modern monitors with a standardized format, which can be read over the display cable [1]. It essentially enables "plug and play" for monitors, as opposed to the bad old days when you needed either an .INF file to install in Windows or the supported sync ranges to type into your XFree86 configuration if you wanted anything above 640x480. It contains data like the vendor, model number, physical screen dimensions, and supported display modes. There is a field for a serial number; parent is proposing using this to identify individual monitors. However, errors in EDID are quite common, e.g. using incorrect units for the physical dimensions or leaving placeholder/OEM data for vendor and model number that don't match the finished product's advertised vendor or model number. I imagine there are a lot of monitors out there with bogus EDID serial numbers, since it isn't important for getting the display modes right.

[1] https://en.wikipedia.org/wiki/Extended_Display_Identificatio...


Yes the pain working with that is real. One of the funnier instances of edid messup was when we discovered there is at least one dell screen model out there that has DEL as their vendor string. They misspelled their company name.

The usual problems were missing mode lines that were actually supported (and needed) and incorrect dimensions.


OP is suggesting Apple should have stored the layout based on the EDID serials.


The 30 Hz/60 Hz refresh roulette is something I have, too.


> For a ~$6k setup, it’s awful. It’d be awful for $1k. About 10 years ago, I had a Dell D830 running Fedora and later, early versions of Gnome shell. The dock powered the laptop and a mouse, keyboard, external drive, and two monitors. It was a single (albeit proprietary) connector, and the dock was fairly expensive (~$200 IIRC). It worked damn great, nearly every time, even with sleep States in Linux on a laptop 10 years ago.

And what did we learn here?


> And what did we learn here?

That I should pay Apple $10k for my next MBP so these features can work?


I dunno what you're whining about, I've had multi monitor setups on several mac laptops and they've always worked fine. Same on my desktop. Same on my work desktop. It's almost always more reliable than doing it on Windows.


Ditto as I commented above, but I’ve heard these reports from lots of technical and smart people leading me to believe there’s a real reason. As others above have mentioned it sounds like laziness on the part of peripheral vendors, but it’s a bit surprising Apple doesn’t have a workaround that makes this not crop up.


> it’s a bit surprising Apple doesn’t have a workaround that makes this not crop up.

If the problem is in fact lazy device manufacturers botching the identification data then I'm not sure how a third party could effectively work around the issue.

If I make two devices that report identical ID information there's no practical way for the OS to tell them apart.


Literally all of my issues revolve around display issues. I honestly don’t understand how developers at Apple don’t fix these things... If I am getting up/down from my desk throughout the day and plugging in and unplugging a monitor, sometimes my screen is just _stuck_ black and requires a hard reboot. Other times, it shows the login screen on both monitors. Yet other times, it decides it didn’t actually detect that monitor.

Most folks on my team have the same deal to differing amounts. We don’t use any special software.


That's not a new problem. My 2012 rMBP w/ 10.11 has the same issue. OS X has always had issues with multiple displays.


Do you get kernel panics too if you don't follow the right sequence of "first wake from sleep then connect external monitor"? I can't believe this is still so broken.

2017 13" TouchBar MBP running 10.12.6


For some reason those panics stopped happening to me after a year or so (2014 15" running 10.10.5 here). Now it just panics if I plug a DP cable in cockeyed.


I am always baffled by these reports. I’ve had nothing but success with external monitors on my last three MacBooks. What is your setup out of curiosity? Monitor brand, connection type, cable brand etc.

I’m assuming I’m an outlier because I see this in almost every thread that mentions MacBooks and I am real curious how it manifests.

Edit: reading further comments it sounds like possibly the monitors use the same edids? Wonder if you can check that easily.

I’ve almost always used Dell monitors fwiw.


> I am always baffled by these reports.

You shouldn't be. This is an actual GIF from the Ars Technica review of High Sierra: https://cdn.arstechnica.net/wp-content/uploads/2017/09/Sep-2...


There's a certain amount of selection bias.

I've had over the years a 2005 PowerBook G4, three 2010ish MacbookPros, a 2014 MacBook Air and a 2016 MacBook Pro. All have occasionally hung on a blank screen after lid open. The latest routinely forgets Dell screens (two identical ones at once) and I have to replug them.

Otherwise, no issues (except random kernel panics). Like brains, I'm astonished that they work at all, let alone work so well. Of course I'd like something bulletproof, but Apple doesn't seem to be able to make one.


The LG 4/5k monitors Apple had in their own stores didn't work half the time


I do think the monitors are partly to blame. They seem to want to go into low power mode right before the computer sends a signal.


Sounds fun! Every day is workstation roulette!


Don’t forget it randomly picking what device to send audio to. Every day it’s a lottery for me. Is it the left or right monitor? The built in dac in my elgato dock? Or maybe it’s the laptop in clamshell mode.

Then there’s the case of it getting an ip from the docks ethernet.

And all of that is assuming the thing even wakes up.


Same here, but 2x LG 5K monitors.

1. Stopped using clamshell mode (laptop lid closed) and that has helped.

2. Downloaded the LG firmware update for the monitors. Updated both monitors, also seems much more reliable.

(10 day uptime, up from about the usual 12 hours)


Or how about it crashes your computer and you'll have to hold down the power button ?


It’s weird, I keep hearing things like this but have encountered literally no issues with it on my mid-2015 MacBook Pro... Is there some particular hardware that it’s dodgy on or am I just lucky?


This is anecdata but…

I have to reboot my late 2013 MacBook Pro a few times a week, when it freezes or becomes partially unresponsive in weird ways. Previous to High Sierra, I tried to remember to reboot it every few months, mostly out of superstition.

I plan to upgrade to a 2015 model when this one dies, for the reasons mentioned elsewhere in this thread and here https://blog.pinboard.in/2016/10/benjamin_button_reviews_the...


Can confirm that review, my first MBP was the 2015 model with El Capitan, it has been nothing but sheer joy for me. I don't even have to deal with temptation because everything that has come out since has been terrible.


Try the NVIDIA Web Driver for video. The new Metal-based window manager doesn't play well with something about the stock video driver. NVIDIA's is faster for me.


I have more than the usual anecdata, curious when it becomes data. :)

Personally, 6 Macs upgraded and I had one problem that was caused by crappy vendor software which assumed port 8080 would be available for them to use.

In the office we finally got the okay to update everyone and 100s of devs are doing fine. There was a homebrew issue that, if you hit it, required a reinstall, but that wasn’t so bad.

High Sierra finally kills Python 2.6 in favor of 2.7, and I had an app fail that embedded Python 2.6 pyc files as an obfuscation — 2.7 wouldn’t run them. I worked with that developer to debug and got a new build out — they were receiving many reports of High Sierra issues that were all caused by that.

Pretty typical OS upgrade, other than the early stuff — but there’s a reason you don’t upgrade until at least .1 is out!


Who takes the time to write a blog post about how they bought a Mac recently and everything is just going fine as expected? Not many people.

Look at it this way: there are literally dozens of complaints posted in this thread. But Apple sold 5.39 million Macs in the most recent quarter.

You're always going to hear more from people who are unhappy, than the quietly satisfied.


There is also those who silently live with the problems because it is too much work to get a resolution. I no longer use a mac but a few years back there used to be wifi issues with my recent ibook where when you open it up, the wifi doesn't find the network half the time. It took month for any fixes even though this is Apple developed hardware and OS and people were getting no answers on their help website. Eventually it got fixed in one of their updates only to come back on yet another update before finally being resolved.


More anecdata - late 2013 15” Haswell + GeForce 750m rMBP, have literally (and I really truly mean this) had 0 issues. Not a single kernel panic or notable userland software glitch. The only thing that glitches occasionally is plugging into an external monitor, but this has happened on every release since Mavericks (what this laptop shipped with IIRC)


I’ve had no problems with my MacBooks outside of what I’d expect (periodic smc firmware resets), but my wife had nothing but problems with the two nvidia MacBooks she owned.

It was a shame because they were, on the surface, basically the best thing since sliced bread for dgpu on Apple hardware, so you must have gotten a gem :)

Or she just got a few lemons out of the millions manufactured.


My suspicion is that it is to do with the set of third party software and peripherals used.

In my case, that’s a 2nd gen 15” touchbar MacBook Pro, a 15” 2015 MacBook Pro, and a Skylake i7 hackintosh tower. I use Asus and Dell monitors and stick to first party or Mac exclusive third party software wherever possible. All three give me little or no trouble with high sierra... reboots are infrequent (1-2 per month, sometimes less) and are usually done to install updates or boot into windows.


It's more the 2012–14 MBPs... I think Apple just couldn't be bothered to give their QA staff anything that was made before the past couple of years...

I feel like my 2014 rMBP is silently yelping "What have you done to me?" every day.


Try the NVIDIA Web Driver for video. The new Metal-based window manager doesn't play well with something about the stock video driver. NVIDIA's is faster for me.


Don't know myself.

MBPr with the strip myself, High Sierra latest, absolutely no problems. Connect it to external mechanical keyboard and Dell monitor when at work, also have external audio cards and MIDI gear. And many external HDs.

Where exactly are all those issues?


I assume a lot of variables come into play, probably what you use, how you use it, and where you use it.

But it's funny you say that, because that dichotomy seems to define High Sierra; it's either a total mess, or totally forgettable.


I actually didn't have any problems with WinodwsME (seriously). It ran about the same as Windows98 ... I mean still buggy and would crash once a week, but that was the entire Win9x era.


98se2 was solid as a rock for me :)


After updating to High Sierra, OmniGraffle, an app that I use fairly regularly, stopped working. The only available solution was to buy a $50 update to the latest version. At least Omni Group should be happy about macOS 10.13.

In Omni's view, my software was hopelessly outdated... But in reality it worked just fine for my needs even though it wasn't the latest version. I had bought the license in November 2009, so exactly eight years before Apple broke it for good.

Microsoft would never break compatibility with a major Windows app from only eight years ago. I've accepted Apple's hostility towards older 3rd party software in exchange for a sense of forward momentum in the OS, but that pact seems to be history now with High Sierra largely feeling like a step back.


> Microsoft would never break compatibility with a major Windows app from only eight years ago.

Microsoft is nearly pathalogic in this. I still run some programs I bought for DOS today on Windows 10, because they're COM files. Still works the same.

Apple has always broken things on the other hand, but with the promise of a stable system. Which, having upgraded to High Sierra, is long gone. This feels almost like the early days of Linux. I've got WiFi breakage, graphics driver breakage, occasional filesystem breakage, and several really concerning CVEs that have come over the last few months.

Never thought I'd see the day when I call Windows more stable than a *nix.


Here[1] is a short video showing the version by version upgrade of Windows 1.01 to Windows 8. Notice how essentially all user programs, system preferences (e.g., theme), and files are preserved throughout the entire process. The guy also runs an installed copy of Wolfenstein 3D on almost every version to demonstrate that it still works.

The first time I saw it made me respect and appreciate the stability Windows has provided over the decades...

Note: there are similar videos that show the exact same process on actual hardware.

[1]: https://www.youtube.com/watch?v=vOzkNtnaS04


\* Compatibility. Which is truly awesome.

But stability means it doesn't crash, or have driver issues or basically interfere with a user trying to use it.


I got to say... I have a couple MacBooks I use at home, but my work PC is a 2012 era AMD CPU with 32GB RAM and a GTX650 Ti with four random secondhand displays connected to it, running Windows 7, and I never have any issues with it.


Well, indeed MS wouldn't break that compatibility.

But the Mac way, to which you sign in when you get a Mac, isn't running 8 year old software forever. It's updating things every few years, and breaking stuff going forward, not being burdened with backwards compatibility forever.

In exchange you get a more uniform and comprehensive experience, with software that adapts to new features faster.


> In exchange you get a more uniform and comprehensive experience, with software that adapts to new features faster.

Do you know how old the version of Bash that ships with High Sierra is?


I know. I also know the reason (GPLv3, and companies avoiding it like the plague). I also know how to install the latest bash with brew and set it as my default shell (takes all of 30 seconds).

OS X is not about the version of bash it includes -- that's a historically hidden behind the UI tool for us geeks, not what's intended for the everyday Mac user to use (and the UI had been the priority there ever since the original Mac).

Lastly, I've used bona fide certified UNIX machines from vendors like IBM, HP and SUN that had much worse core userland -- awfully limited BSD versions of GNU tools, much older stuff, etc. So there's that too.


I'm not a developer, but I knew developers were previously content with Macbooks usually running OSX or linux. Given so much unhappiness with OSX, and the inability to run linux on the USB-C macbook pros, what are developers generally doing?

The Apple hardware is nice, but it generally means you have to use OSX. Windows WSL is pretty convenient, but then you have to use Windows. Linux is, of course, great, but what hardware?


Well this is purely anecdotal, but at least 70% of the developers, IT workers, and researchers that I know are using prior generation MacBook Pro hardware (the ones without the touch bar or terrible keyboard, and with ports) and are intentionally using the older OSX versions.

I know several people who traded in 2016/2017 models for 2015 models because it's literally an upgrade; get an older machine and you suddenly have usable ports, a keyboard that works, and an escape key, what a luxury. I also know a fair amount of people getting Thinkpad and even Surfaces.


An especially unfortunate truth is that many developers have grown so afraid of workflow-breaking bugs that they're willing to risk security flaws just so they can get work done.

The things I would do (and pay) for 2010-Apple tier hardware built for CentOS in both laptop and phone form factors...


> The things I would do (and pay) for 2010-Apple tier hardware built for CentOS

Isn’t it remarkable this still doesn’t exist? Build quality still does not compare in the PC world.

Yet, I’m on a MBA3,2 (late 2010) and it’s great. Still running Yosemite. It’s got 4gb ram but handles running 10 JVMs with modest heaps for development no problem. Stable and runs for months without issue. I’d love more RAM and more cores but none of these computers seem worth it.

I considered buying a 2015 MPB the other month only to learn that Apple glues the battery to the top case. So to replace the battery you need to replace the whole top case. Insane.


> Isn’t it remarkable this still doesn’t exist? Build quality still does not compare in the PC world.

Razer comes close, though I wouldn't buy them because of the compensating-for-something #branding on the machines. The Alienware 13 I can personally vouch for as being better than my last MacBook Pro for sure. And the keyboard is like an old ThinkPad's but not quite so heavy. Still not a fan of the alien logo, but at least it's tiny.


> Yet, I’m on a MBA3,2 (late 2010) and it’s great. Still running Yosemite. It’s got 4gb ram but handles running 10 JVMs with modest heaps for development no problem. Stable and runs for months without issue. I’d love more RAM and more cores but none of these computers seem worth it.

I have the same hardware but unfortunately my experience is the exact opposite. It’s barely usable. I literally can’t browse JavaScript-heavy websites (e.g. Gmail).

I wonder if having kept up with the OS releases has anything to do with it (I’m on High Sierra).


I'm using an MBP5,1 (2 years older than yours) as my main personal machine still, it's quite comparable to your MBA3,1, (P8600 vs SL9600), both Core2Duo roughly same clock speed, same RAM and bus speed etc, but different GPUs, also I added an x25-M SSD back when that was "new" :P it's lasted well...

I'll admit some of the really heavy websites are crap, but that's mostly because the websites are crap and wont be smooth on anything than the latest i7. Generally I have no problem with gmail or google maps which are reasonably heavy.

Depending on what version of OS X you are using that might be the culprit, OS X is a pretty heavy OS, especially post 10.6. For me 10.6 was o-k but changing to linux and a tiling window manager freed up a lot of resources, I never end up paging out with web browsing even though it's only 4GiB, additionally I find graphics are faster on linux but that may be GPU and driver specific due to Apple's underclocking tactics. If I hadn't switched I would probably have ditched this hardware by now.

Alternatively the GPU difference between these machines may explain web browsing speed considering how much stuff is hardware accelerated these days.

I think this era of hardware is pretty linux friendly if you wanted to try it, you don't have dual GPU which is the biggest pain for linux too... there is no issue with mac EFI from grub anymore so should be zero configuration install for you, at the most you might have to manually select the proprietary drivers for your GPU for reasonable performance.


Weird! I’m sorry to hear that. My fan starts whirring on video/graphics heavy sites, like landing pages with huge videos or overwrought effects. But sites like Gmail are totally fine. Maybe it is the OS updates? I almost pulled the trigger and updated the other day but I’ll hold off now with this story. Gah, that is frustrating.


“Insanely great”


> An especially unfortunate truth is that many developers have grown so afraid of workflow-breaking bugs that they're willing to risk security flaws just so they can get work done.

Can't blame them. Ultimately, it's productivity that pays the bills.


It's not always even bugs though. I remember back in ~2013 OS X updates would regularly break my Postgres setup because Apple like to muck around with it on every release. Postgres.app solved that, but the disregard for developer issues has been somewhat annoying since a longer time. And then you have a IDE like XCode that feels many years behind any JetBrains IDE lacking basic refactoring tools.


> 2010-Apple tier hardware built for CentOS

Librem? https://puri.sm


Have you tried a Dell XPS 9350?


Also anecdotal, but this is also what we're doing at our company too. Most of us have the 2015 era Macbook pros, and run a split between 10.9 and 10.10. Others that aren't using that gen are using Thinkpads with full keyboards on them with Ubuntu.

It's not ideal, but our environment is rock solid for what we need, and newer software that we need still supports us just fine.


> but at least 70% of the developers, IT workers, and researchers that I know are using prior generation MacBook Pro hardware

You can have my 2015 MBP when you pry it from my cold dead hands.

But yes. We recently sold off a bunch of hardware resulting from workforce reductions; I clued a friend in on this and he was begging me to nab him a 2015 MBP. The obsession is real, it is the last truly great Macbook in a while, in my own opinion.


The 2015 MBP is the one I don’t use any more but keep anyway because I know it will never fail me. I also stay one version behind on it, so it’s the only system I have using Sierra still. But my main system now is an iMac pro and you would have to send in troops to separate me from it. Although I still can’t stand their crappy keyboards, regardless of color.


I believe those are also the last linux friendly ones so you know that once you either get too sick of the newer OS X whatevercat or the version your are stuck on stops receiving security updates you still have useful hardware.


My employer has a shiny new 2017 MBP for me that I have yet to touch because there are SO MANY issues with both the hardware (the need for four different dongles and no escape key) to the security issues to the fact a lot of core software still doesn't work solidly. I'm sticking with the 2015 MBP they gave me before.


This. Although I didn't redowngrade to the non touch bar, I literally hate it and stare enviously at my colleagues running Dell and Lenovo Linux laptops with 32GB of RAM, etc. I am still hoping for Apple to get its act together but I am done with the crappy non Pro "Pro" consumer junk laptops.


There will almost certainly be 32GB MBPs this year when Cannon Lake shows up. Assuming it isn’t delayed again, naturally.


I would not bet big money on new CPUs arriving as previously scheduled. There is about to be a major re-think on CPU architecture with security in mind - or so I would hope.


I've been lucky that with my past three jobs, I've been able to use Linux naively. One place just had custom desktops, another HP workstations and the last I got a Dell laptop. I ran openSUSE on the HP, and Gentoo on the other two.

One of the shops was an open source shop, so all 150+ of us ran some Linux distro (mostly Ubuntu/Debain, some Arch, some Mint .. I was the only Gentoo) except for the graphics people who got Windows or Macs. But I was surprised at the other two shops, there were very few Linux users. There were some here and there around the company, but I'd usually be the only one on my team who ran Linux naively with Windows in a VM.


I'm in a similar scenario, we basically get freedom of choice over OS, with the caveat of being responsible for our own installs of course. This works well small scale with competent devs you can trust to take care of their own OS (we help each other out if needed obviously, but the point is that you become fairly self sufficient in that respect, no "IT dep" needed).

I know there are reasons this can't happen at large corporations such as security... more to do with contracts and culpability than measurable security improvement AFAIK.

I can't quite imagine _not_ having control over my work environment in this way in a job anymore and do greatly appreciate it. It's not all about familiarity, there are many objective reasons using certain proprietary OS can be a significant drain and inefficient, even more so for development work. This freedom would be a significant factor in considering a new job for me, even over pay because it's a big part of my "happiness" and "sanity" factor in the job.


> I've been able to use Linux naively

Hehe


And arguably better battery life.


> Given so much unhappiness with OSX

Unhappy people post a LOT more and the internet loves a good pile-on. Apple has had plenty of embarrassing QA lapses but the average developer is fine, and it’s not like everything else is perfect. The Linux users I work with spend more time on sysadmin work than the Mac users by a hefty margin, although there’s a good argument that software compatibility is worth that overhead.


I have to say (and of course this just another anecdote) none of the developers I know (including myself!) are happy any more with OSX. In particular, no-one wants the new MB Pros. We put up with the gradual culling of ports for the sake of the things we liked (OSX's great 3rd party GUI software ecosystem, plus enough unixyness underneath, and the great hardware). But removing 1/6 of the keyboard for the sake of a feeble "innovation" has been a cut too far, and now we're gazing with envy at machines with real keyboards and an actual diversity of ports.

The only developers I know who are replacing their pre-touchbar machines right now with new ones are those making much of their living from iOS / OS X. Many (like me) are kind of hanging on, because it's a pain to change OSs, lose much of the software we love etc. I'm still undecided, but I think my current (ageing) MB Pro is going to be my last.

> The Linux users I work with spend more time on sysadmin work than the Mac users by a hefty margin

Yep. I shifted from Linux to OS X years ago because I was sick of tech fiddling and wanted to have a machine that needed no hand-holding and let me get on with my work.

There seem to be no good options.


> The Linux users I work with spend more time on sysadmin work than the Mac users by a hefty margin

This is why I use LTS Ubuntu. I don't do any sysadmin work at all. Every couple of weeks or so a dialog opens up saying there are updates, and I click OK. Everything "just works".


I switched to Linux running on ThinkPads (between work and my personal laptop I have a Carbon, T470s and a T450). I have generally found ThinkPad hardware is very compatible with Linux.


Thinkpads.

I moved from Windows 7 to Debian 8 a year or so ago and have never looked back (except for the Adobe Suite...)

I use an OSX Macbook Pro for my 9-5 and have been avoiding High Sierra forever - it's just a slap in the face.

At least I know my Debian 8 box will almost always operate as expected - even through a distribution release. I need stability from my system, and there's just no other option anymore.


Apple hardware was nice. Not anymore. I had to have the keyboard replaced on my touchbar model. And the 8 Gig memory limit is silly. Thanks to an external keyboard and vagrant I get to spend most of my time on linux with a good keyboard, so it is tolerable for now.

I would dump it for a linux laptop, but I work remote so every type of video chat/screen share tool needs to work. Most of them don't work on linux. My computer choice is dictated by chat tools, which is kind of sad.


Hangouts, Slack, Zoom.us, Appear.in (even Skype iirc) work fine with Debian.


I had a company like that. I just ran Windows in a VM and connected it to the company domain controller. While at work, I used x11vnc on one display, tightvnc on the Windows VM on another monitor and would use that with WebEx if I needed to share my screen.

At home, I ran the same setup but I used my gaming/windows laptop for my VNC client and WebEx. I could never get the Linux Cisco VPN client to work, so my work laptop just made an openvpn connection to my flat (fibre connections are nice).


I use Linux as a daily driver, and currently have zero issues using Hangouts and Hipchat's built in videochat, and only minor issues with Skype, for a data point...


> And the 8 Gig memory limit is silly.

You mean 16, not 8.


And no native Evernote client :(


That is indeed unfortunate but the windows version runs decently on Wine. It's a pain to figure out the exact versions of Evernote and Wine that will work + the fixes (currently I'm on Evernote 6.4 and Wine 2.1 using PlayOnLinux). In the end some annoying bugs linger, but for the most part I still prefer it to Evernote for Mac.


Still?


FreeBSD 12(current) on a 2017 XPS 13 has been great for me. Funny thing is I never imagined buying a dell laptop. I was going to purchase a new mac, but the touch bar killed that for me. So far so good with the XPS 13.


I can at least confirm for you that you made the right choice. Today I went to use the touchbar to adjust the volume on a conference call I was in and ended up hitting the contextual "home" button, taking me away from the page (and call) entirely. I guess I should have memorized which invisible part of the touchbar corresponded to what better. I miss hardware buttons.


Did that end the call? Most conferencing tools I use are happy if you shift away. I have to say I hardly ever use the volume buttons on my Mac and I use it 10-14h a day.


How much of a hassle was it to set it up? Are all peripherals working?


To get a working install it was very easy. I just grabbed the latest FreeBSD 12 snapshot iso and put it on a usb drive and installed like normal. Everything but the wifi works. So I'm using a usb wifi dongle. You need the new (experimental) drm-next code and drivers for suspend and resume support. If your comfortable with a unix command line environment you can get a working dev environment in a couple hours.

So besides the wifi everything is working and iirc someone has been working on a driver for the builtin wifi card and plans on merging it into 12 sometime this year.

But so far I'm loving it, I have ZFS, Dtrace, Sublime3, Firefox57 and xterm, which is an ideal dev environment for me.


I’m toying with converting mine from arch to FBSD on the same machine — where did you get the drm-next bits? Is it from Ports or a branch on /usr/src? I bought a dongle to try it out but somehow ended up with one of the few unsupported ones... which one are you using?


  sudo pkg install drm-next-kmod 
The command above should get you the required drivers and stuff. Then follow the directions that are displayed after running that command.

The USB wifi dongle is a TP-LINK TL-WN725N. I found it at my local Microcenter.


Nice! What usb dongle are you using for Wi-Fi, if you don't mind me asking?


No problem, I'm using a TP-LINK TL-WN725N. I found it at my local Microcenter for under $20. Otherwise look at this list of supported wifi NICS [1].

[1]: https://www.freebsd.org/releases/11.0R/hardware.html#wlan


> Linux is, of course, great, but what hardware?

We are in 2018, not in 1995. You can easily find good desktop and laptop that run fine under pretty much all major distros.

Especially if you compare to MBP & the like, meaning you have no budget issues...


Apart from high DPI displays which is not great at the moment.


> Linux is, of course, great, but what hardware?

Plenty of threads on this forum (and others) to choose from when it comes to choosing hardware for Linux. The answers are often similar: Thinkpad T4xx, Dell XPS 13/15, System76. I don't see HP mentioned that often but I have a bunch of (older) HP laptops running here and there which do just fine, just stay away from their consumer products and focus on (used) business hardware. Purism is cropping up here and there as well, mostly due to their active stance against Intel ME.

In other words it is not that hard to find good hardware to run Linux on. Given the state of MacOS these days the experience is likely to be less trouble-prone than that of those staying on (or hopping on) the Apple platform.


> inability to run linux on the USB-C macbook pros

What have you tried?

I have a MacBookPro14,1 (2017) and had a bit more of a problem than expected triple-booting Windows, Linux, and High Sierra, but I got it working.

My main issue seemed to be that nvme wasn't supported for many of the Linux installers. I think it was a newer Fedora I finally got working. I may have had to manually mount the drive before running the installer (I cant remember). I haven't used it as my primary OS, but everything seemed to be working (hi-dpi, wifi, trackpad).

I think like any other new hardware, it'll take a couple years before Linux works out of the box.


I have made the switch to Ubuntu for Windows on my personal projects. If you aren't trying to run daemons, it is a perfect workstation environment.


I recently bought a ThinkPad X1 Carbon. I'm pretty happy with it.


A useful filter when buying laptops is "does this laptop have the service manual accessible to mortals?"

Handily filters down to 2 companies: Dell and Lenovo. I am unsure about Librem and PogoLinux.


HP has service manuals available too, including their consumer models.

Oddly they come with the disclaimer "IMPORTANT! This document is intended for HP authorized service providers only." which makes you wonder why it's available to download from the support website.


I came for the incredibly useful touchbar and stayed for the cascade of un-exit-able menus that occur when you try to combine the awesome powers of the fingerprint reader and keychain.app


The one where the escape key used to be? Fuck I hate that thing...


As a sysadmin at my company, I have regular questions from my users (C++ developers) about upgrading from Sierra to High Sierra due to Apple's invasive and unavoidable notifications. After the root-exploit balls-up, I forced the two people who'd upgraded back onto plain Sierra. But I know at some point it'll stop giving us the option to ignore.

I am beyond flabbergasted at how much these idiots have managed to break between Sierra and HS. They must literally be high when they say 'this is good, let's ship it'. The OS is abysmally unfinished, and with each of these stupid failures we're going past Vista levels of unfinished. These are tools I use in my daily workflow (thankfully I use Linux instead), and other developers in my company use these tools. At this point, upgrading to HS would be a massive step backwards for us.

I literally cannot believe how sh*t MacOS has become. It's a cesspit of broken Unix with a shiny iOS layer covering it. I absolutely loved Snow Leopard and cannot believe it has evolved (or mutated) into this. I'm beyond pleading and into exasperation.

APPLE, WILL YOU MAKE AN OS WE CAN ACTUALLY USE?!


This is why we can’t have nice things--random reports of things not working are pretty meaningless without context.

Yes, it’s fodder to continue to dump on Apple—I get that. But it certainly doesn’t help solve the issue, assuming there’s really an issue in the first place.

For the record, I’ve installed htop many dozens of times using Homebrew on a variety of Mac hardware and operating systems over the years and have never experienced an issues.

After many years of Mac IT tech support, I can tell you: it’s usually what’s not being reported that's the problem.

Was this a clean install or was it upgraded from a previous version of macOS?

Are there 3rd-party kernel extensions installed? Pretty common with antivirus software and hardware drivers.

When’s the last time Disk Repair was run? I’ve seen corrupted/damaged hard drives cause machines to kernel panic or hang.

Did htop ever work correctly? If so, when did that change?

I’d be pretty surprised to hear a new iMac that was unboxed and had Homebrew and htop installed would crash. Actually, if that were the case, they could boot into the recovery partition, format and reinstall.

I clean-installed macOS High Sierra 10.13 the day it was released on a 2009 iMac and have been running the public betas and haven’t had any issues since.

I’m running macOS High Sierra 10.13.3, build 17D29a. I’ve installed 157 command line and 71 GUI applications using Homebrew…

As someone mentioned, it’s only the problems that are part of some narrative that make news; people like me don’t post “Hey—I’m a web developer and I’ve been using High Sierra on a nearly 10 year-old iMac with zero issues, even when running beta versions”.

This doesn’t excuse Apple at all for the bugs and vulnerabilities that have been well-publicized; but lets not make it seem worse than it actually is.

Apple has already starting releasing mitigations for Meltdown and Spectre, like many OS vendors are doing.

I run a FreeBSD server on Digital Ocean; here’s their statement:

FreeBSD was made aware of the problems in late December 2017. We're working with CPU vendors and the published papers on these attacks to mitigate them on FreeBSD. Due to the fundamental nature of the attacks, no estimate is yet available for the publication date of patches.

This is pretty underwhelming, especially given the severity of these vulnerabilities.


> This is why we can’t have nice things--random reports of things not working are pretty meaningless without context

No they are not. A lot of the same reports with same combination of htop + iTerm should prompt the company to investigate. At least when they are selling $2k "premium" machines.

> I’ve installed htop many dozens of times using Homebrew on a variety of Mac hardware

Neither did I except since I upgraded to High Sierra, happened within an hour after.

> it’s usually what’s not being reported that's the problem.

Usually is. So who from Apple is investigating what is not being reported. It seems like after enough people seeing the issue, they'd prioritize it a bit.

> Did htop ever work correctly? If so, when did that change?

Even if htop doesn't work correctly, and is complete garbage it doesn't install kernel extension and doesn't run with admin privileges so it shouldn't freeze the machine or render it unusable which is seems to do.

> it’s only the problems that are part of some narrative that make news;

The corollary there is that to make the news and get the attention, would need to build a narrative. In many cases with large companies with lots of consumers the best way to solve a problem is to make a big stink on the social media. It's unfortunate but because it's effective, that's what people use.

> but lets not make it seem worse than it actually is.

For people who see their expensive machines which used to work, stop working is pretty bad. Had these been some no-name $150 cheapo chromebooks there wouldn't be such a big deal.

And this issue might not be as serious, but you do recall the "root party" vulnerability, just a month or so ago? That was actually pretty bad, wouldn't you say? And at some point the effect is cumulative. One thing after another starts to get to people. So people will let off steam online about it.

> This is pretty underwhelming, especially given the severity of these vulnerabilities.

One is a non-profit foundation and another is a company with almost $800B in capital. There is a slight difference and different expectations there as well.


> doesn't run with admin privileges

To run correctly on macOS and display all processes htop needs to be root via sudo/setuid.


No they are not. A lot of the same reports with same combination of htop + iTerm should prompt the company to investigate.

You don’t know whether or not they’ve investigated this issue or not.

There’s a decent chance they have, which might explain why I haven’t had this problem because I’m running a beta of 10.13.3 and nobody else on the GitHub thread is as far as I can see.

I also haven’t seen evidence of bug reports to Apple; that might’ve helped: https://developer.apple.com/bug-reporting/


> You don’t know whether or not they’ve investigated this issue or not.

True, because the company we're discussing isn't exactly the champion of communication when it comes to things like that. A simple reply from an Apple developer that it is being looked into, that it got linked to an internal KB, would've been suffice.


One is a non-profit foundation and another is a company with almost $800B in capital. There is a slight difference and different expectations there as well.

1. You’re confusing Apple’s market capitalization (around $890 billion) with it’s cash on hand + marketable securities , which is only $268 billion: https://www.macrumors.com/2017/11/02/earnings-4q-2017/

Regardless, it’s not relevant to this discussion since we know all of these companies who make tens of billions each quarter have bugs and security issues.

Given your logic, there’s no excuse for Intel creating vulnerable processors going back to 1997, with literally hundreds of millions of computers that can be compromised and hacked. It wouldn’t take more than a few percent of these machines to be hacked to potentially cause havoc all over the world.

The point being: money doesn’t mean you can’t screw up; being a non-profit doesn’t let you off the hook. It’s not crazy to expect the FreeBSD team to say more than “we’ll get back to you” regarding fixes. At least the Linux guys are dealing: http://www.zdnet.com/article/major-linux-redesign-in-the-wor...

FreeBSD also doesn’t have over 1 billion vulnerable devices in the wild either; their threat model is tiny compared to what Apple and the rest of Intel’s OEM’s have to deal with.

I know, I know: Apple is always held to a different standard; they should give equal weight to an incorrect compiler flag for htop as they would for preventing more than 1 billion of their smartphones, tablets and computers from being hacked.


> I know: Apple is always held to a different standard;

That's exactly what I was getting to. They should be held to a different standard.

https://forums.freebsd.org/threads/63985/

They were notified in late December, before the holidays. Intel knew for 6 months at least. So even there some vendors had more time than others.


Even if htop doesn't work correctly, and is complete garbage it doesn't install kernel extension and doesn't run with admin privileges so it shouldn't freeze the machine or render it unusable which is seems to do.

Turns out it doesn’t, but it’s not unusual for people to make these types of unsubstantiated claims; turns out it’s a build issue that conflicts with High Sierra.

The Homebrew team updated the formula so it doesn’t try to install on High Sierra until they can fix it.


Speaking of making unsubstantiated claims: this post is totally untrue. The formula was not updated with an OS requirement due to some misconstrued build issue; installation on High Sierra was blocked 6 hours ago, as noted in the very issue thread that this HN topic is about [0], explicitly because the freezes and crashes are very real and reproducible.

[0]: https://github.com/hishamhm/htop/issues/682#issuecomment-355...


> Turns out it doesn’t, but it’s not unusual for people to make these types of unsubstantiated claims

What are the unsubstantiated claims? That these people including me are lying and have made this crash up? Do you really believe that?

> turns out it’s a build issue that conflicts with High Sierra.

Thanks for checking, but what is the build issue is exactly?

The original point was that unless htop installs kernel extensions or runs as admin it shouldn't break the rest of the system. So there might be an htop issue but there is most likely a macOS issue as well.


Whatever the system issue was is fixed for some and certainly for me on 10.13.2 or newer when running htop.


> Was this a clean install or was it upgraded from a previous version of macOS?

You do realize, don't you, that if this is a relevant question I already have cause to complain? It's been the future for some time now, give us reliable upgrades already.


You do realize, don't you, that if this is a relevant question I already have cause to complain? It's been the future for some time now, give us reliable upgrades already.

Yes, it’s the future, but as I mentioned, the Mac I’m running High Sierra on is nearly 10 years-old—it shipped with a version of Snow Leopard, MacOS x 10.6.

That was 7 major operating system releases ago—pretty sure nobody at Apple tested upgrading from an operating system that hasn’t been supported by Apple in many years, not to mention all of the permutations of hardware, kernel extensions, file systems, 3rd-party software installs, etc.

This is the double-edged sword: no matter what Apple does, they’re wrong. If they don’t support older machines with the latest operating system, then it’s a conspiracy to force its users to upgrade.

If they do support older machines and can’t guarantee perfect upgrades from every version operating system they’ve ever shipped, regardless of the condition of the user’s machine and how well (or not) the machine has been maintained, they’re also wrong.


I'm not saying it's easy, I'm saying they should have solved it anyway. For a company that prides itself on (and charges for) high quality user friendly products, reliable software upgrades should be foundational. The double-edged sword argument only makes sense if you assume it's impossible, when in fact it's only the preventable (given Apple's control) accumulation of cruft that makes it expensive. Upgrades certainly would have made a better investment than removing headphone jacks, or adding the touch bar, or almost anything that went into High Sierra.

Yeah, I'm an idealist. Sue me.


> If they do support older machines and can’t guarantee perfect upgrades from every version operating system they’ve ever shipped, regardless of the condition of the user’s machine and how well (or not) the machine has been maintained, they’re also wrong.

Yes, a $50 billion company that doesn't run massive amounts of automated regression tests on one of its flagship products is Wrong. Come on, Microsoft manages to support absurd levels of backwards compatibility (see the discussion upthread about upgrading from Windows 1.0 to Windows 10), what's Apple's excuse?


Turns out there’s some kind of build issue with htop; the Homebrew formula has been modified to not install on High Sierra.

Here’s what I got when attempting to upgrade; for the record, htop still works fine for me:

    ==> Upgrading 1 outdated package, with result:
    htop 2.0.2_1
    htop: This formula either does not compile or function as expected on macOS versions newer than Sierra due to an upstream incompatibility.
    Error: An unsatisfied requirement failed this build.


Yeah, that's done as a result of this linked bug report. Its mentioned in the bug report somewhere at the bottom.


You can silence the error by running "brew pin htop". Remember to "brew unpin htop" when/if the bug is finally fixed.


You should blame whoever gave advance notice to Apple and allowed them to patch the issue, but didn't do the same for FreeBSD.


Sadly, my Powerbeats headphones need me to upgrade to it in order to connect with Bluetooth. (???)


You mean having to install a MFT of additional software ala homebrew etc instead of just installing your Linux distribution of choice on a cheaper more powerful system was ever a good choice for a workstation?


I also, am seriously considering downgrading from High Sierra for various stability problems... i hope Apple fixes whatever went wrong.

Did they shift too much developers to iOS and forgot about macOS?


It seems like every day I read a new reason not to make the update from sierra.


  1. By creating and closing process very rapidly, one could easily reproduce
  this problem. I used something like `while true; do sh -c "exit 0"; done`.
The entire OS dying to rapidly spawned/exited processes permits a pretty nasty DoS. Good thing that nobody ever uses macs as servers or shared machines.


Author of comment here: Doing this alone doesn’t do any harm unless running htop in background, also this only makes the bug happen more quickly. If one has Activity Monitor running prior to crash, it’s still possible to kill htop with it and would make OS useable again.


Just tried it on Ubuntu (Windows) for fun. 5 minutes later the fans have come on but no freezing yet.


Geez; does this mean it could do the same when compiling a large codebase with Make?


Dunno; I don't have any apple devices. Might ask a coworker to try to repro it on monday. :P

(I suspect that recursive makes do enough compute in between invocations to slow down and avoid this bug. If the OS was dying in everyday compile jobs it'd be much bigger news.)

EDIT: I think that I might've misread this - it might be that spawning processes pushes htop through codepaths that can cause the bug. So it's more like "a malicious app could kill your system the instant you open htop". In which case, yes, a recursive make while you have htop would would do it. Rather less significant.


When I read the headline I didn't think I had run into this bug. Now I remember my OS freezing twice when quickly opening and closing some chrome tabs.

Could possibly be the same issue.


I’ve been disappointed with Apple’s response to kernel bugs for quite a while. Closing a large number of ptys at once caused a panic for years. In high Sierra, there’s this, statfs hangs, and proc_pidinfo hangs.


Reading the bug report:

1. Increasing htop's refresh rate (-d 0.1), and having a tight loop that spawns processes quickly, tends to reproduce fairly reliably. 2. Running htop as root seems to work around the problem.

I was able to reproduce the crash with (1), and not able to with (2) after running htop for several minutes. My first thought was that it didn't have to with running htop as root, but rather running htop as a different user as the one spawning processes. So I repeated (1) except that I ran htop as myself, but the spawning loop as nobody, and the crash went away.

I wonder if the bug's in the resource limit checks (RLIMIT_NOFILE, RLIMIT_NPROC), and if increasing the limits would delay / mitigate the crash. But I don't feel like crashing my machine anymore tonight to find out...


Some nice digging there, thanks for the info!


On Mac OS 10.4 (tiger) when I left Activity Monitor going in the background for a few minutes and returned to it, sometimes it would all of a sudden refresh the values super super fast with historical data (appearing to catch up on the iterations that it did not display.)

On 10.6, when I ran Windows in a virtual machine while it was doing windows update, and left it running in the background for a few minutes and returned to it, sometimes I would get 60fps in the machine for a few seconds and it updated very very fast, and then it went back to 15fps (seemingly getting more CPU power.)

On 10.7 (if I recall) if I left Activity Monitor running in the background for a few minutes and returned to it, I would get nonsensical values for the CPU usage for a few seconds (e.g. 398423.34%) but then it would return to normal. Perhaps this error has been accumulating over the years.


Oooh! I recently had a super high percentage event, it was because of a bunch of IO that happened right when I woke the Mac up, and I guess the system monitor dropped a few samples and nearly divided by zero.


This still happens for me on OS X Sierra sometimes.


Out of curiosity, is this bug related to CPU cache? I don't know really anything about how CPU's work internally, however. It just "feels like" it's applying a bunch of instructions very quickly but ignoring the internal clock if that makes any sense.

For example, the CPU percentage counter in Activity Monitor was around 3,000,000. If there was a loop to increment the CPU usage counter, it sounds like it incremented it so fast that the thread that reset it didn't run since it wasn't scheduled to run (as it only runs every few seconds.) Again, this is just pure speculation and if anyone knows otherwise please let me know!


I very doubt it's anything to do with the cache, and that isn't an uneducated guess. But I've very doubted things before, and they turned out to be the case anyway, so why would you listen to me?

My assumption was/is that it refreshes stuff on a timer. On each timeout it puts a refresh event on some kind of queue... and then somehow manages to fuck it up, and ends up having a bunch of these events to deal with all at once, rather than handling each one as it comes in with a sensible gap between each one. Well, sometimes it happens, and don't ask me how I know.

But I have to say that you usually don't release it like that :(


Somewhat tangential but if htop is unusable, I’ve always liked Glances:

https://nicolargo.github.io/glances/

Though it sounds like htop is a symptom not the problem.


Glances seems great but every time I go to use it it’s a pain to install. Python packaging... yuck!


People are missing the point-- it's not an htop problem since a simple console app running as a user should not be able to do this at all.

This gives me the willies. Methinks another OSX security bug is incoming.


Yea, if I had a Mac box, I'd be looking at that call stack the author posted and fuzzing the hell out of those functions. There's a DoS bug bounty in there somewhere.


Only if Apple invited you to that competition (https://www.macworld.com/article/3104671/security/apple-anno...). Others only get credits.


This is exactly why I cringe at these seemingly minor issues - sure htop breaks, it can be worked around, but the root cause is more fundamental; an unprivileged user should not be able to crash the whole OS just by reading the process table, no matter how invasive.

It makes me think what they have decided to change between Sierra and HS to cause this sort of failure. This is pretty fundamental and implies some tinkering deep in the kernel, potentially without fully understanding the repercussions. With the number of security holes that can be easily opened by naive tweaks to low-level code, I think we're right to be nervous about what they've done to cause this.


Why was this reported to htop instead of Apple? Most of these people "me too"ing don't seem to realize that this isn't htop's problem...


Yea, there's very little indication anyone here has tried to do any serious debugging or realizing this should be a security vulnerability report. The author did eventually post a stack-trace to the actual call that's failing.


Good Lord, how dare you imply that is actually Apple that made something wrong? /s


This seems like it might be a lead:

> Apple completely rewrote the window server in Metal 2 for High Sierra and seems to have done a very poor job of it.

I've run into tons of those issues with my 2012 MBP. Slight tangent that might be useful to someone.

If you experience jerkiness, lagginess, slow display updates, just general slow feeling, one thing to try is the NVIDIA Web Driver video drivers, rather than the stock ones.

It resolved all of the slowness for me, maybe just applies to slightly older macs but made a big difference.


Since Mac OS X Snow Leopard (10.6) the quality has declined release by release.


> "Since Mac OS X Snow Leopard (10.6) the quality has declined release by release."

This may be: I'd love to see a comprehensive, in-depth article that digs into this in a quantitative manner. Do you have one at hand? If so, it might make a good HN submission. Just posting this statement at the top level of a post is arguably off-topic and baity. If you could post it on any submission that touches on Mac OS X, it's not really on-topic.


I've had a similair problem with Transmission for years. Doing more then a few transfers locks up transmission and then the entire machine. Issuing a halt at the terminal sometimes will reboot the machine, but moist of the time it just hangs.

I would switch but to what? Windows/Linux are just as user hostile.


Qbittorrent is quite nice, in terms of switching torrent clients at least.


I'd say OS X is hostile now - after doing maintenance on a 2015 Macbook over the holidays.


I'd agree. It's just relative.


Yup. I hit the same bug, just a few hours ago. Was compiling something and had htop running, then everything froze. I assumed it was the build and after a hard reboot started compiling again but this time without the htop and it finished fine.


I’m still looking for a thing I _can_ leave running, without High Sierra freezing on me..


Linux?

:)


From the symptoms, it looks like it’s starving the computer of memory. It doesn’t look like anyone’s debugged this, so once I get back to my computer I’ll take a look and get it reported to Apple.


I can't edit anymore, so I'm going to reply: I did a bit more research on this issue and it's not actually a memory bug at all. In reality there appears to be an infinite loop in calls to task_for_pid: https://github.com/hishamhm/htop/issues/682#issuecomment-355.... I'm still not sure who's doing the wrong thing, htop or Apple, but it's clear that freezing up the system is always incorrect behavior.


Kinda sounds like an htop bug that shouldn't be able to freeze the system because, y'know, not letting any process freeze the system is part of the OS's job. Of course, it might also be an htop bug that's triggered by Darwin handing it bad data, so there could be blame to share;)


After a Quick Reading of the issues it reminds me of an old linux Kernel bug.

https://groups.google.com/forum/m/#!topic/mechanical-sympath...


I have an old MacBook Pro I use as a gaming machine (and my wife uses as temporary machine after our late dog (for reasons unrelated to her demise) urinated on her new XPS13.

After upgrading to High Sierra, we both got so sick of the instability that we're installing it with Xubuntu this evening, bringing it in line with the rest of our 'home fleet' (ThinkPads w/ either FreeBSD or Xubuntu).


Oh crazy I was wondering if my system was totally messed up for quite a few weeks now. Thought it had to do with my homebrewed version of htop but didn’t have time to investigate further :/


Ran htop for a while on the latest public beta of High Sierra 10.13.3 last night and didn’t have any issues.


According to the bug thread, the bug doesn't always happen but there are some things that can be done to reliably trigger it.

https://github.com/hishamhm/htop/issues/682#issuecomment-341...


Wow. I used to update my iOS and macOS every year, but decided not to do so in 2017. A stroke of luck?


My 2015 13 inch MBP is still on Sierra. Should I indefinitely hold off upgrading to High Sierra?


Yes. Stay away.

I downgraded my mac back to Sierra from High Sierra and don't regret it at all.


+1. I used to upgrade to each new release of Mac OS X in the first few days. That policy worked for many years, but my experience with High Sierra was so bad that I downgraded and never looked back.


I’ve been finding the root cause of this a month ago, but don’t know where to look due to lacking of experience in kernel debugging. If anyone here could lend me a hand, please send me an email at (es at evsfy dot com)


why keeping the root cause for yourself? Try to mention it at least here or write on the htop bug list page what you know. It would help us all I guess.


I did post my findings on the GitHub issue :-)


Okay for now htop is disabled for High Sierra:

==> Upgrading 1 outdated packages, with result: htop 2.0.2_1 htop: This formula either does not compile or function as expected on macOS


Not quite: many people already have 2.0.2 (which is buggy) installed. Short of uninstalling htop the issue still remains.


High Sierra STILL has a bunch of issues running Unity too.


macOS has gone down hill since it heyday in the 2000s, very noticeably so since it became known as 'macOS'. For a long time, it was the best nix you could run. The horror stories surrounding High Sierra might finally break my bad habit of commercial OS and make me go back to Linux or one the BSDs.


Leaving htop running (best in background in a screen session on some server) is such a waste of CPU resources. I've already seen that multiple times and always wondered if the people don't recognize by looking at htop's output how much CPU resources it wastes.


Htop is pretty leaky on any system.


Yes, with a small difference that this does not happen on a Linux box.


I always wondered why people bother with htop on OS X when the built-in Activity Monitor.app is so nice already.

I only use htop on remote boxes myself.


I have always wondered why people bother with non-htop apps, when htop is so nice already :)

But more seriously, one of the big advantages of htop is that it is extremely lightweight, which helps a lot when you have other processes that are eating up all your CPU and RAM.


I think htop is a tolerable replacement for top but personally I find it just okay, not "nice".

The main reason I prefer Activity Monitor is that it has more features, like viewing open files by process, the energy usage pane and network activity on top of CPU and memory. I tend to prefer GUI apps any time I'm interacting with the interface a lot, like to sort, filter, drill down, etc.


htop allows viewing open files by process (press 'l'), as well as sorting (F6 or click column header) and filtering (F4 or '/').


This has less to do with your choice of process monitors and more to do with the potential security issue that allows a userspace app to halt a running system.


Personally the first thing I do on a macOS install is remove pretty much every single Apple app I can. The quality of Apple's software is simply too low to take the risk of relying on it. Anybody who has had to fix their relatives iPhoto or iTunes would agree.


If your confidence is that low, why use MacOS at all?


Despite the flaws it still provides a more consistent experience than Linux (especially on laptops). I just wish Apple put as much love into MacOS as they do into iOS. Also there is a lot of proprietary software like the Adobe suite + Music production software that is only available on MacOS or Windows.


You should try KDE Neon stable :)

It's the best linux experience I've had, I got used to call it "macosx, the linux way". It's stable, with a well thought of interface and softwares that just fit together, yet it's customizable to the extreme and will allow you to create your workflow instead of forcing one upon you.

Not that it will help with the unported softwares you're used to use, though (although, inkscape, blender and ardour are a thing, nowadays).


> more consistent experience than Linux (especially on laptops)

One one line of laptops, maybe. Run Linux on a single line of laptops that officially support it (I suggest XPS), making this an apples-apples comparison, and Darwin doesn't look as competitive anymore. Or cut the other way and look at Hackintoshes:)

> proprietary software

Yeah, this I agree with; WINE helps but sometimes you're stuck needing Darwin/NT.


Not the op, but until recently I felt like while Apple's apps were going downhill, its OS stayed very good, both in terms of user friendliness and stability. With High Sierra, any trust I had was utterly shattered.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: