I stopped caring about OS reviews after virtually every Yosemite (10.10.0) review failed to mention the drop in performance and battery life and the discoveryd and Wifi problems last year. These were not isolated issues and a lot of people suffered them, I'd say in a much higher proportion than any first release of OS X since 10.0. At least the Ars Technica people are smart and recommend people to wait until 10.11.1 if they can (although last year Yosemite 10.10.1 didn't solve anything).
I have to chuckle when I see some reviews saying "UPDATE _RIGHT NOW_! THIS IS THE BEST OS EVER!". They said the same last year, what a coincidence. Personally, I feel that Yosemite 10.10.5 is very stable now, and I'll wait at least until 10.11.2 to upgrade, no matter how many times people tell me how wonderful El Capitan is.
Odd. I'm guessing you both use them fairly similarly?
I'm not saying Yosemite is perfect; I'm sure there were problems with it at the start. To me, the original person I responded to seemed to kind of just make a blanket statement and I wanted to make sure it was known that there are counter-examples (based on anecdotal evidence anyway).
That would be a good question. They're identically spec'd machines of roughly the same age with batteries that are within 1% full charge capacity. They re both on the same wifi, Bluetooth is off. Display brightness varies, but neither is cranked, I'm not sure what else hardware wise could be seriously different.
The biggest difference is the os.
I will concede that usage varies, hers hits Facebook, mine Twitter. And I run virtualbox but she doesn't. We're both running safari, chrome and mail.
I have a 2014 MacBook Pro and had the Yosemite Wifi issues. And looking at the Apple forums a ton of other people had them too. It has become better (from totally unusable) but Wifi still cuts out from time to time.
I wonder if these problems were already apparent during the test phase.
I never had any WiFi problems with Yosemite. The thing is, there are millions of people using macs and even if 0.1% have problems, forums will be swarmed by them.
I can say that I definitely experienced and experiencing laggy animations compared to mavericks. Launchpad animation is like 3-5 FPS and is very noticeable. But probably many people just don't care about that.
I have no issues. Until I enable my Bluetooth mouse. Easily solved by using the 5ghz band but really. An apple mouse fucks up my connection from my apple laptop to my apple access point. That's really unacceptable.
I too had WiFi trouble when bluetooth was enabled. The commands below fixed it for me (it might be that only one was effective but I've had no trouble since Feb 7). YMMV.
Once Apple ditched DiscoverD, my wifi and bluetooth woes disappeared instantly. I had a problem where my connection speed suddenly would drop to no more that about 7k as second through wifi which required zapping the PRAM on my 2013 MacBook Pro until a week or so the connection would start exhibiting the same problems. mDNSresponder seems to work like a charm in Yosemite.
That's true, but also kind of the point. If reviewers can't figure out a review process that illustrates the issues, how useful are they, really, other than a look at features?
I don't want to speak for spike021, but it is possible this is an edge case that is specific to carlosrg. If it was a localized issue specific to how the OS was installed or how that computers environment was set up, what can a reviewer do? There are as many programming environments as there are programmers. If a review can give as much insight as possible about how the subject changed from the previous version, as well as technical details I consider it a good review. If they themselves, have problems, it would be awesome if they include them as well as any other qualitative data. On balance, there will be subtle differences across hardware and software that can not be addressed in granular detail. The best way to get information is just to read several reviews, and see if any of those people had edgecases.
edit: wiredfool pointed this out above, as I was writing this. Having purchased 2 equally specked computers at the same time and experiencing differing results. After upgrading from MAvericks, I personally experienced significantly better battery life. I have edited launch daemons and blocked some outgoing connections which caused coreaudiod to go into an infinite loop using >80% of the cpu. This KILLED my battery and performance. After fixing this issue, both of those metrics were amazing.
This was a bug for which we know the root cause and for which a patch was actually developed. If it wasn't affecting you, I will argue it is more likely that you just have a high tolerance for these kinds of conflicts.
hmmm. I was referring to CPU and battery above but I looked into it a bit further. Also, connectivity is HORRIBLE on my iphone iOS 9.1 so thank you. I don't know how to confirm the wi-fi issues independently here.
I have gamecenter.app deleted, airdrop and bluetooth disabled and many of the sharing and push notification services are disabled. I have fairly decent connectivity and not much seems to change after disabling awdl0, a speedtest seems pretty similar.
I am now really interested. Is there a way I can run a better speedtest, and also check if these issues you describe are happening because it sounds as if it is sporadic. Also, is there a way to use something other than bonjour/mdnsresponder via wi-fi? I could not find anything which is why I still use it unfortunately.
edit: Also, almost forgot.
> This was a bug for which we know the root cause and for which a patch was actually developed.
What? I have seen issues surrounding coreaudiod but none of them are referenced in your post. The post actually explained the iOS connectivity issues I am having but doesn't seem related to a glitch in the core services daemon, which seems to affect some users, and was almost certainly self-inflicted. Now, I want to test my connectivity to confirm, but it seems alright most of the time.
Interesting post. I had come across wifried in the past but didn't personally feel a need to use it. You definitely have a fair point that it could very well be a case of "out of sight, out of mind." Maybe not as noticeable for some people, especially since it's also possibly related to older hardware not supporting full AirDrop.
So basically, are you expecting the reviewer to spend hours and hours of QA testing to try to find bug that the OEM didn't find during the pre-release test?
It seems obvious to me that a technical review will be pretty limited in term of "stability" evaluation. They will run a few tests, but it can't cover all the configurations. Especially things like the wifi issue which were mostly triggered depending on the network environment (what is your access point? What other devices are connected? etc.)
I would imagine the reviewer could have done some investigative journalism: put a call out for people having problems with the beta, collected stories, and set up configurations to try to reproduce those. Not just trusting what people say, but not going all the way to QA, either.
It's incredibly difficult to figure out what portion of users experience a certain issue, especially for a very new (or even unreleased) OS, and especially for things like WiFi performance and battery life, which most users don't measure rigorously and thus may be imagining or exaggerating the problem.
I've got the 10.11 GM running on my 2012 Mac Mini (Core i7) with spinning rust, and it has made such a huge improvement over 10.10 in terms of snappiness and how fast the entire OS feels.
It is a world of difference, to the point that I am happily using my Mac Mini for stuff again, whereas before I would avoid it as much as possible and instead reach for my MacBook Pro Retina to get anything done.
Now, there was a trend with OSX releases that, until about Lion or so, every new release was faster and more optimized than the previous one. We are used to expect just the opposite, specially from competing operating systems.
I have not seen signs of that trend in the more recent releases.
Windows is getting faster on the same hardware since Vista (7 was faster than Vista, 8 was faster than 7 and 10 is faster than 8). Linux is difficult to judge here because of the variety of DEs, but it's not really slowing down with time either.
Newer iOS versions are slower on old hardware than older versions, and this is also more or less true for Windows Phone and Android.
Yes, I've regretted upgrading to Yosemite when 10.10.1 came out... First thing, Filevault managed to crash and corrupt my hard disk. Had to restore from my backups (not a huge issue but a waste of time).
Then, I cannot connect to either iphone or my wife iphone's hotspot without turning the wifi off first and then turning it on again.
I have the same issue on both a macbook pro and a macbook (but didn't have that issue before Yosemite).
And in term of performance, my macbook pro is nowhere as performant even with the latest Yosemite update as it was before. I also have noticed a strange issue. On yosemite, whenever the battery reaches 5%, everything slows to a crawl...
I feel that the last really stable OS from Apple was Snow Leopard
Same here, Yosemite is the new Windows Vista ! I'm finally stable enough with 10.10.4 after many kernel panics, not gonna rush it to upgrade. I'll wait for 10.11.1.
I'm just praying that my wifi doesn't start disconnecting again every 1 minute...seriously wifi is one of the most basic things to have in a laptop. Why is it so hard for Apple to get this right?
Here's to hoping Bluetooth on my Macbook won't stop working every time I put the computer to sleep. Then hanging the entire system when I try to toggle it off/on.
Oh man, I thought I was the only one. I've searched a bunch about this online but couldn't find anything. This mainly happens when I'm connected for an audio speaker. Weirdly, connecting it to my Bluetooth keyboard after the fact sometimes helps?
Oh man, the buggy Bluetooth stack in 10.10 is a huge source of frustration for me. At first I thought it was my peripherals, but they work without any trouble within EFI or Windows on the same hardware. I even tried poking at the stack using Bluetooth Explorer (part of Xcode), to no avail.
Well, to answer my own question - nope, still having problems with Bluetooth in 10.11. Normally, I wouldn't complain about something so basic, but it's really, really annoying when something that I use every day (like my BT mouse) only works sporadically in Mac OS X.
Edit: I've tried creating a new network location, resetting PRAM, resetting SMC, deleting Bluetooth PAN, re-installing the OS. Nothing seems to work reliably, and I don't know what else to try at this point.
I own a number of different machines running the same software and one of these machines just has flaky wifi. Random disconnects or stalls during large downloads. No software update is going to fix that machine.
If you have a router that supports 2.4ghz and 5ghz networks, try using the 2.4ghz one and give each network its own SSID. That's what worked for me, but it's obviously a shot in the dark.
This. Create "MySSID 5" and "MySSID 2.4" (or however), and then rank the two in priority order, 5 over 2.4. Then the OS will prefer the (usually less crowded, but shorter range) 5GHz network and fall back as needed.
I used to have the two on the same SSID, but I found my adapters (or the OS) would gain an 'affinity' for 2.4 as the stronger signal and never use 5.
Why did you have to phrase it like that? I don't feel it benefits discussion. It doesn't feel like the HN way.
If you had done an inquiry into the matter, linked to a stack exchange discussion on this bug persisting or being fixed that would've been a lot more useful. Maybe some technological quirks would surface that we could learn from. It's such a shame submissions like these, which the author clearly has spent a lot of time on, will get derided with snark like yours.
Now jkimmel continued in this vain; your comment has anchored the discussion in what i feel is a negative way.
Because it is frustrating. Judging by my post getting positive votes as opposed to negative ones, it appears I am not alone in this. Incremental improvements to the Mail app are lovely, but I (and many others) do not use it. I use wi-fi all the time.
If you Google "macbook wake up wifi does not work" there are plenty of complaints, a couple of half-hearted solutions, but not much else. I didn't link to any of them because it isn't worth anyone's time to have to read them.
Yes, I am technically minded. As such, I could debug this further and further, but:
1) I can only go so far. This problem requires Apple to issue a fix, and they're definitely aware of it.
2) it's a little like asking a car mechanic to do some car repairs in his spare time for fun. My Macbook is a tool, I want it to just work.
Yeah, I totally get where you're coming from. I'm not asking for you to debug and fix the problem. I get that this is deep-seeded, and the ball is on Apple's court.
I just don't want HN to be a place of one-liners. In your original post you were complaining without adding any substance. Sorry if I sound harsh, it just really irks me. Maybe I should've left it at a down-vote, this back-and-forth is not particularly fruitful either.
I just don't want HN to be a place of one-liners. In your original post you were complaining without adding any substance.
I guess we can agree to disagree - I don't really know what useful substance I could provide. Plenty of people experience the same thing, and I don't think posts should be wordy just for the sake of not being one-liners.
Because it is frustrating to debug a closed system.
I am all for starting a stack exchange discussion when working on things that built communally and not locked down. However, when you are paying Apple for a product, it is frustrating to deal with things being broken and on top of that you have to circumvent their restrictions.
I fear that this is one of the worse OSX updates. I've seen a lot of issues in the GM. For example my USB audio device (an amplifier) is no longer working. The Wifi connectivity issues are not fixed. The Airdrop connection between my MacBook and the iPad Air does not work. A Lisp system did not start because Apple removed some font protocol.
Stuff like Screen split between two apps are non-intuitive.
There is lots of good stuff, but Apple created also new problems or failed to fix really annoying old problems (like Wifi connectivity issues, the buggy and unreliable Finder, the annoying slowness of iTunes with larger music libraries, ...).
I think the review overstates the problems that developers might have with System Integrity Protection. It seems to me that if an application developer has to disable SIP, then that app developer is probably doing something wrong. Better to accept SIP as good for everyone, like app sandboxing.
I don't understand why developers don't just follow Apple's own Xcode example: ship a sandboxed app that, on first startup, offers to run the (elevated) installer for an accompanying OS-integration package. Or does Apple not let anyone but themselves do that?
Whoa, thank you so much for bringing PlistBuddy to my attention. I feel as much joy as when I learned about `/usr/libexec/java_home`. Maybe I should just start poking around in /usr/libexec and see what other awesome toys are hiding in there.
If your application is not a game, or it does not fit the model of "here be a document, type on it", chances are the sandbox won't allow it. Actually, even if you do follow that model: see BBEdit and Coda.
What about Postbox? iStatsMenu? Heck, even Alfred. Or Parallels, Microsoft Office, the Adobe suite? These cannot be sandboxed.
Frankly, I could come up with more applications if I did some research, but why bother. Nearly everything that's interesting that's not a game will be crippled or just unavailable. And, for games, Steam is better in every way.
EDIT: By "cannot be sandboxed", I mean that they cannot be sandboxed under Apple's current draconian sandboxing rules.
WJoy is my go-to example of an app broken by SIP: it's effectively a third-party driver to change the properties of the Bluetooth stack (and do some extra higher-protocol-layer autonegotiation magic) so that Nintendo Bluetooth devices will connect to OSX. Absolutely needs a Kext to do what it does.
I really don't know why a custom kext is needed. I had great success without any of that. http://blog.mcohen.me/other-stuff/miitransfer/ I also made it so that when I pressed a wiimote button it would simulate pressing a kVK_ code. Maybe it's so that it gets recongnized as a gamepad directly? I remember that being kind of odd.
I'm curious how much Metal matters but it seems like they went from GPU to CPU in Yosemite and then back to the GPU.
The review sorta skimmed on Metal but reported that it only worked on a Mac Pro 2013, which lead me to wonder why my 2008 Mac Pro couldn't run it with a GeForce 680 GTX, turns out it can.
http://netkas.org/?p=1405
What's the recommendation if you're using command/shell utilities on upgrading vs. clean install from Yosemite to El Capitan? I usually do a clean install, but would an upgrade be advisable?
I've done upgrades since 10.1 was released without issues. If you've followed Apple's filesystem layout and haven't altered system files or rely on some exotic kernel extensions, you'll be fine.
The one annoyance I've encountered is that the installer will move everything in /usr/local out of the way before doing the upgrade and moving it back. If you have a large Homebrew install, it will be much faster to move the directory first and restore it later rather than waiting for the file-by-file copy to complete, even on when using SSDs.
I only did upgrades on my old MBP too, and always found it fine. I went from Leopard to Snow Leopard to Lion to Mountain Lion to Mavericks. Nothing serious went wrong at any point (though I did have to reinstall or update the odd program), and there was no noticeable deterioration in performance over time.
There was the odd infuriating bug (I'm not convinced it's tested as thoroughly as Windows is...), but overall it was fairly trouble free and the smooth upgrade process played a part in my decision to buy another. It certainly wasn't the price, at any rate.
> The one annoyance I've encountered is that the installer will move everything in /usr/local out of the way before doing the upgrade and moving it back.
This can be a significant point. I thought my computer was bricked when I did one of the previous upgrades (maybe Mavericks?) and the upgrade took something like 14 hours (with no change in the progress bar).
It's definitely mind-boggling just how inefficient it is – feels like 1-byte writes followed by a full fsync based on how it can keep an SSD capable of 10K ios busy for a period measured in hours.
It's potentially contradictory: the alternative is upgrading your OS and find pretty much everything where you left it, instead of having to reinstall everything. I.e. life is too short to spend time reinstalling everything.
Surely, it's an opportunity to do some cleanup of unused stuff, but otherwise, I haven't had a lot of problems upgrading OS X. (nothing memorable at least)
Too add to this, on my personal MBP I have the same home directory that I originally created in OS 10.2. I've used migration assistant, swapped/cloned drives, and in-place upgraded for the past 12 years. My work MBP is similar for the past 6 years.
I've had 2 or 3 issues where old config was the root cause, but those are few and far between. I might be an outlier, but it's always just "worked for me".
I've upgraded my Macbook Air since Lion without issue. There's no reason to expect trouble coming from the new security features as homebrew/macports as well as sane make scripts default to installing outside system directories anyway. In the worst case, you'll just have to reinstall the offending applications.
Of course you still do a backup before in case the universe blows up.
Homebrew installing outside system directories hasn't protected it in the past and the directory/root lockdown isn't the only change, remember the tun/tap CF that got dropped in their lap? The 10.9 to 10.10 transition broke a lot of things and more than a few were caused by the security changes. Do not trust homebrew to work properly on 10.11 until they can find & stomp on all the bugs Apple has graciously handed them, it's not a safe bet at all and more than a few network apps (for example) do need root at some points.
Such a great tool. It can more often than not correctly suggest resolutions to problems with homebrew. It’s a level of helpfulness to which many dev tools can aspire.
Is there any way I can get this today without being a "certified apple developer" or whatever the language is. I have to reformat anyway and really don't want to reinstall yosemite and then have my settings ovcerrun on the upgrade in 2 days.
While that took minimal effort on your part, I want you to know that it was really really helpful. Thanks a lot. Is this new or something? I couldn't find it a few weeks ago when the GM was released?
It's a shame he's retired from doing them, given the grievous collapse in the quality of the writing at Ars Technica, but he's got a life to live. I suppose.
have you paired any iphone over bluetooth., try to remove it and turn off the bluetooth. check if you still getting that issue., I created a local account and turned off bluetooth., havent seen it for a while now.
And another version release updating apps I don't use, ushering in new interfaces nobody wants, while leaving core infrastructure (e.g. containerization, a better launchd interface for an amazing init system, an ancient filesystem, allowing things like rendering contexts across process bounderies, etc) leave me scratching my head. I feel like they're trying to convince me to abandon my tools because they have been messing with them for next to no reason since snow leopard. (exception: I love the app store and sandboxing.)
While it's true that this release doesn't touch your pet issues, I'm left scratching my head as to why you expect that Apple will ever address some of those issues.
Containerization is unlikely to ever get a big push from Apple -- whatever half-hearted designs they had on the server space are long-since abandoned, and it's clear they favor far less heavyweight security measures for end-user apps.
"Rendering Contexts across process boundaries" doesn't even sound like it has a good use-case for any direction Apple's been working on.
> "Rendering Contexts across process boundaries" doesn't even sound like it has a good use-case for any direction Apple's been working on.
Well web sandboxing for one, and running opengl through your terminal for another.
Containerization seems like a no-brainer to me since a) sandboxing naturally follows, b) it's a developer's heaven in terms of deployment, development, and testing, c) mac os x is now the only major OS in existence without any containerization capabilities controlable by anyone but apple.
And these aren't just pet issues, no developer I know (except perhaps iOS developers, for obvious reasons) welcome them. Even the non-technical users I know see them as "pretty" PITAes.
And it's not that I don't get why they do it, I just wonder about how they view the *nix-ey developers who use the platform. I'm sure there must be plenty at the company itself.
> Well web sandboxing for one, and running opengl through your terminal for another.
Sharing OpenGL contexts is about the opposite of sandboxing. It's unsafe.
And you can run OpenGL apps from your terminal just fine?
> Containerization seems like a no-brainer to me since a) sandboxing naturally follows,
But it's a sandboxing that isn't useful on a Desktop machine. A mail app which sees a completely different container filesystem than my word processor does not improve my life.
> b) it's a developer's heaven in terms of deployment, development, and testing
Except it's not, because OS X on servers is dead and gone, so what I'd really need is a container that works on my server, aka a Linux server aka I need a VM locally because it's not like OS X is ever going to be LXC compatible.
A container I can only use on my dev box isn't worth beans.
> c) mac os x is now the only major OS in existence without any containerization capabilities controlable by anyone but apple.
I get that Docker has grabbed a lot of hype by reinventing jails to be venture capital compatible, but honestly, for OS X it makes absolutely no sense whatsoever.
Apple does continue to expose new infrastructure to third-party developers. For example, Yosemite added Hypervisor.framework, which made xhyve (http://www.pagetable.com/?p=831) feasible. xhyve has the potential to be very useful for tools like Vagrant and Docker Machine.
You can't share a rendering context between processes, but you can share an IOSurface between processes, and bind that to an OpenGL texture, which you can either draw using, or render to (by binding it to a framebuffer).
<rant>
Ugh. Apple software quality has been going downhill so fast lately I loathe seeing the "Software Update Available" notification.
If I'm lucky everything will still work, and it will merely scramble the interface and hide a bunch of stuff.
Unfortunately, my expectation is that on top of the interface scrambling, a bunch of stuff that's worked fine for ages will be broken in some stupid way I can't even imagine how it could happen, like it will start duplicating playlists on my phone again. Oh, and I'm guessing it will try to shove at least one unwanted, half-assed Apple service down my throat, integrate it with everything, and force me to search online for 20 minutes looking for how to turn it off, only to find out there's either no way to opt out, or the option is hidden in some obscure, unexpected location in System Preferences, or via some command line BS.
Every new release from Apple I get one step closer to switching back to Linux. Love their hardware, but their software people need to get their shit together.
</rant>
So, don't update? I'm still running 10.8 on my 2008 Macbook and I only install security updates. If the new version doesn't have any features that you must have, why update?
Because Apple uses a version ratchet to force you to upgrade.
I essentially use only one app on my iPhone. It's a mobile banking app. On a regular basis, my mobile banking app stops working and requires me to upgrade. Sooner or later, to upgrade the banking app, I have to upgrade my iOS, which eventually forces me to upgrade iTunes, which eventually forces me to upgrade OS X.
Western civilization has been in decline ever since Snow Leopard was deprecated. I hung to that as long as I possibly could, but even I eventually had to upgrade to Mavericks. Mavericks is tolerable, but the writing on the wall is clear: to use Apple products, you will eventually have to surrender to the software update.
To address your point, not updating really isn't an option. Sooner or later some software I use will make use of a library or framework from a new version of the OS, and then I'll be stuck either not updating it, or not being able to get my work done.
And if nothing else, next time I buy hardware I'll be getting the current version of iOs or OSX whether I like it or not.
And for the guy saying, "Don't sync your phone with iTunes," is that even a real suggestion? What happened to Apple products "just working"?
Really, it shouldn't be too much to ask that the richest company in the world do a little QA on their software so it's not buggy as hell, and not force "features" like Apple Music on everybody for the sole purpose of squeezing more money out of them.
I have to chuckle when I see some reviews saying "UPDATE _RIGHT NOW_! THIS IS THE BEST OS EVER!". They said the same last year, what a coincidence. Personally, I feel that Yosemite 10.10.5 is very stable now, and I'll wait at least until 10.11.2 to upgrade, no matter how many times people tell me how wonderful El Capitan is.