A lot of people seem to dislike the way Windows install updates, and Microsoft seems to be doing something about it. In current test builds, it’s improving the update experience.
Have you ever had to stop what you were doing, or wait for your computer to boot up because the device updated at the wrong time? We heard you, and to alleviate this pain, if you have an update pending we’ve updated our reboot logic to use a new system that is more adaptive and proactive. We trained a predictive model that can accurately predict when the right time to restart the device is. Meaning, that we will not only check if you are currently using your device before we restart, but we will also try to predict if you had just left the device to grab a cup of coffee and return shortly after.
I’ve never had any issues with Windows updates – they just install automatically overnight, long after I went to sleep, and hours before I wake up and start using my PC again.
Windows does an absolutely terrible job restoring after a restart. Unsaved work is typically lost. The trouble with Microsoft’s solution here is that you don’t know what you lost when it comes back up. It’s not like “hey, notepad had an unsaved document opened, it’s gone now”. Instead you are presented with an empty document or no notepad window at all. Hilariously, you do often see a balloon or other indicator that “hey, we installed updates while you were away, isn’t that great?”
This is the big issue. Some of us like to leave work up and on so that the next time we sit down it’s waiting to be worked at without having to go through an annoying series of clicks and waiting, whether it’s later in the day or overnight (I learned to save before sleeping/locking the screen long ago).
We’ve moved on from the days when people shut down their machines every time they were done; Microsoft does not seem to have noticed this. My Macs have slept beautifully since the late 90’s; Windows hibernate/sleep is dicey, at best. An OS should facilitate work, and something that almost purposefully interrupts it is unacceptable.
I guess you don’t have lightning storms, power outages, cats, or anything else that might cost you your data if something unexpected happens.
Me, I save everything before I walk away from the computer, and can’t remember the last time a Windows update did more harm than good.
You didn’t read what I wrote, did you.
Sure I did.
And what I gleaned is that like many people, you’re impatient, and willing to risk data corruption / loss in order to save a few seconds here and there.
I’m sure that’s not how you see it– I doubt you’ve been using computers since the days when powering off a system without removing the floppies risked zapping blocks (no self-parking heads).
I doubt you’ve taken a lightning strike to your phone modem which sent a power surge into the motherboard and fried the RS232 port (and did some serious damage to the rest of the board, including the attached HDD).
Finally, and most importantly, I doubt you’ve spent much time on computers without solid journaling file systems, so recovering files after a crash tends to “just work”.
But ask yourself this: “What’s your data worth?”
What would it cost to replace those files on your hard drive, right now? Do you have backups? Are they current? Are they on a different system?
An unexpected Windows update should be the least of your worries.
It’s not just open documents.
I regularly leave my machine to crunch data overnight, and I’m not impressed when I come back to it in the morning and have to re-run a 4-hour process because an update has lost the output.
I’ve woken up to find my machines stuck while updating. I’m glad your machine either works really well, or you just enjoy maintaining them, and don’t mention that part – but the update process is anything but lovely.
Yeah, I’ve seen Windows Update just hang in the middle of downloading updates before – no errors, no nothing.
Anyway, I got excited when I read this article, because I thought they were gonna ditch the 6 month update cycle. Did anybody really ask for that? I wish they’d go back to 3-5 year major updates. As is, they keep moving shit around between updates, so every time I Google how to do something in Windows 10, the instructions are for an older build, where the UI is different.
How about a forced update when doing a quick restart?!
That’s absolutely the worst ducking time ever! Fix that!
Here’s a thought…, stop being the god damn update police. Not everyone needs a Windows nanny so allow users to opt-out of your lame ass nanny service. And here’s the great thing about it — it only requires a checkbox! No complex algorithm that tries to “predict” what’s best for hundreds of millions, if not billions of Windows users.
Yeah, who needs security updates?
Oh wait– the people who ask that question. They’re why we can’t have nice things.
Oh sure– you’re smarter than the hackers. You’ve got a firewall on your $40 router / gateway, after all! You know how to browse the web safely, because you’ve got cookies and UAC disabled!
No hacker’s gonna hack your computer!
… because someone probably already has.
Let’s face it– The security in your copy of windows, your router, and your ISP’s router, was all originally designed by people who think turning lights on and off randomly is “good security”.
And the users? Oh, we trained our users not to read UAC prompts. Or SSL warnings. Or to check URL’s when clicking on a link in email. We’ve taught them to accept whatever prompt comes up on the screen, to give whatever access that random person with a foreign accent is asking for, because they’re from “Microsoft Support”.
And let’s not get started about the Internet-of-Badly-Secured-Things– Every clever connected converged hyperfriendly eavesdropping piece of technology you’ve bought in the last 10 years, can be compromised by my cat (He’s a real script kitty)!
And you honestly think forced updates are a bad idea?
You want an OS for technical people, run Linux or *BSD. If the convenience of windows is that important, do the damned updates, scan your computer regularly, and ponder the possibility that if your updates are randomly failing or crashing, your computer has a problem.
Twenty-plus years of trying to keep networked computers safe in the face of unrelenting stupidity has made me somewhat surly about the topic.
I didn’t say security updates are unnecessary, and I didn’t say people shouldn’t install them. I, obviously, did say some people are perfectly capable of managing updates themselves and don’t need some algorithm getting in the way..
Now, … Seeing that you started off your reply with such an absurd comment, I didn’t see the point in bothering to read the rest of it.
The bottom line is this… There are tons of people perfectly capable of maintaining their systems. Yes, there is a segment that needs help and for them Officer Update is probably the best solution. And yes there is a segment that doesn’t need Daddy Microsoft to hold their hand, and for those people, they should have to option to opt-out of the forced automation. There’s no reason both types of users can’t be accommodated. Windows is after-all, a product for everyone, not just a product for children, idiots, and the ignorant.
Edited 2018-07-26 03:41 UTC
Isn’t deferring updates for months still an option on Windows versions (like Pro) that feature the Group Policy Editor or Windows Update for Business? I don’t use that myself, but very brief googling makes me think so…
It’s a pity, you might have agreed with most of it.
You’re right that it was absurd– deliberately so. But it was even more sarcastic than absurd.
It wasn’t actually meant as an attack against you, per se– more the prevailing attitude of people who feel that Microsoft is forcing updates on them for no good reason, when in fact, there is very good reason.
Left their own devices, most people will postpone even the most critical security updates for weeks, or even months– and then be absolutely flabbergasted when their systems are hacked.
I find it entertaining that the most heavily downvoted post I’ve made in 12 years on this forum was the one ranting about the sad state of security and the tendency for people to overrate their own ability to secure their systems.
Kind of justifies the rant, really.
Thing is, “children, idiots, and the ignorant” (as you put it) will probably form the largest portion of those who think they “don’t need Daddy Microsoft to hold their hand” and will promptly disable updating with some instructions found on the web… (remember how it was with UAC? Also, many “IT techs” seem totally unfit for the job, two different ones (one a ~freelancer, the other a computer support company) set a Windows PC for my computer-illiterate parents with disabled updates and automatic login to administrator account…)
I’m not speaking as someone who `thinks` they don’t need Daddy Microsoft to hold their hand, I’m speaking as someone who doesn’t. For the users you’re describing, who I’m not convinced actually are the largest portion, Microsoft (and knowledgeable others in general) needs to do a better job of educating users how to safely user their products.
The solution is not to simply remove the ability for users to manage their own systems, and take away control over hardware they own. For those who insist on clicking everything, opening everything, etc., they will find ways to screw up their system regardless. They will either eventually learn, … or not. But to start from the premise that everyone is a baby and incapable of either learning or knowing how to manage their system is ridiculous.
I have living relatives who are nearing a century of living and are as un-computer-savvy as you can get, who use internet-connected computers, and who don’t have a problems with infection or being hacked because they have learned what’s safe to do and what isn’t. If they can figure it out (with proper teaching) then damn near anyone can.
And I didn’t wrote that you were the one who merely thinks that… But you know, not negligible malware infection rates come from somewhere – and I gave two examples of utter failures of IT ~techs in properly configuring PCs / educating (and possibly with one FB account credentials leak); if even they improperly managed the OS… (a cynic in me might say it was to assure job security / so that the PC in question will return after a while for servicing; but perhaps that’s how I hope they knew what they were doing…)
Edited 2018-07-31 19:53 UTC
Windows Update provides our shop with job security. People stream in after the latest Feature Update to fix their BSOD, orBoot Loop.
A few months back I was on board for bachelor’s thesis of a student, he turned on his win10 laptop with presentation on it and it started to update. We waited for 5-6 minutes, nothing happend (except update percentage went form 0 to something like 15), so I went to get my Linux laptop and we could finally proceed… Needles to say, this was very stressful for the student.
Ever heard of “Active Hours”? I mean, it’s not like you must be a rocket scientist to configure it to suit your needs but, apparently, it’s a lot better to complain all the time about “unexpected updates” than to spend two minutes to use the feature…
RT.
I have never heard of “Active Hours”, I don’t use windows and I don’t care. If I’m to guess what they are for (time when OS do not update/restart?), my computers are ALWAYS in active hours, they update in background automatically, and restart only when I agree for them to do so. I don’t see the need for an “Active Hours” concept, sorry.
Then what about NOT talking about things you don’t know anything about for a change?
And since you said “Linux” and you NEVER reboot your computer: I take it that you never install a kernel upgrade? On Linux Mint I have those coming in at least every other week — and I’m talking about security updates, not the latest and greatest “just because”.
RT.
I guess you NEVER heard about live kernel patching? I restart my computers when I want, every week, or two-three, when I feel like it. And they are updated.
Sounds like little more than a stopgap solution which doesn’t solve the underlying problem: like many parts of windows, instead of getting out of the way and letting one do ones job it is intruding one the user experience in some of the most obnoxious ways possible.
Let’s see:
1) Computer insecure because the user does not install security patches: ripe for zombification and to either send spam or attack other computers —> blame Microsoft
2) Computer secure (relatively speaking) because updates are installed automatically, even forcefully if you ignore “Active Hours, sometimes involving a reboot —> blame Microsoft
So, honest question: what would your “non stopgap” solution be?
RT.
A non-stopgap solution would involve a complete overhaul of the update process: it should be so non-obtrusive that it doesn’t matter when it updates, at the very least for non-intensive tasks like giving a presentation, office work or watching a dvd.
That way you have no need for “active hours”, smart users or spooky ai to try to predict when the user wants to use the computer.
Edited 2018-07-26 22:22 UTC
Lets be honest. Windows is insecure because of the user and because of the OS design. And why on earth should patching disrupt the user. I don’t have to stop anything while my system patches and I only have to reboot when it suits me. Granted, I don’t get the new kernel right away, but the system doesn’t lock me out like Windows running a surprise feature upgrade after reboot while your trying to actually get something important done.
The “active hours” thing is buried in the settings and most non-geek people will not find it, rocket scientists or not.
And anyway, “active hours” does not solve the major problem with Windows Update, which is that I have just finished my work and want to go to sleep, but the computer decided to go into a 2-hour update marathon, doing a vacuum cleaner impression, before it shuts down. I am glad Thom uses the life-hack of leaving the computer on overnight, but I consider it too wasteful and noisy to adopt.
My problems with Windows Updates are:
1) They are slow
2) When building a new system, you can install all updates in one go.
3) In Windows 10, you can’t turn off the dreaded “install at shutdown” option.
Edited 2018-07-26 15:45 UTC
1) Yes, they are slow. Not “2 hours slow”, but I often wonder why I can install a clean OS in a minute, patch it offline in 2 minutes, but need 5 minutes for an online patch and 40 minutes for an upgrade installation
2) I don’t understand why “you CAN install” would be a problem, so I think you mean “CAN’T”, which is incorrect on Windows 10. You only ever get 3 updates (Flash, Platform, Cumulative Update) and maybe some Defender updates that all come at once. Even when you get driver updates it isn’t normal to need multiple rounds of updates anymore. The only exception I could think of is when you postpone major updates and then install 1709->1803 (1 WU) and get the smaller updates later (1 CU) although that normally happens during the installation in the background as well. This is very different from Windows 7 where updating is indeed a many-round process
3) I you have set WU to download and install updates you indeed cannot prevent the updates from finishing installation during shutdown, which makes perfect sense because that part only requires a small amount of time and is needed. However you can set WU to “download only” and now shutting down will not install anything. You might have to use “Hold Shift” or “shutdown /s /t 0”, not sure
I’m not dumb and leave my PC powered on through every night, so Windows installs the updates usually when I want to play games and went to grab a glass of water, or when I’ve had enough and just want a quick reboot back to my Linux desktop.
I set my Windows 10 Pro to update when I want it to.
All you need is to run gpedit.msc, navigate to “Computer Configuration -> Administrative Templates -> Windows Components -> Windows Update -> Configure Automatic Updates”, enable it and set it’s value to 2.
And they say that Linux is unfriendly…
Sorry, I don’t mean to oppose you (this looks like valuable information for those who need it), just to draw a parallel between this and scary things like “paste this in terminal…”
I use both Windows and Linux and can say that both have their own quirks.
W10 Home users don’t have gpedit installed by default but luckily all important files are already present in Windows folder.
For those in need I pasted a simple batch script I found some time ago:
https://pastebin.com/mPWMqpBH
Well, good for you.
You, like most of those who post here is technically literate and can understand all this sort of stuff.
Please spare a thought for the average user. do you really expect them to know all these neat tricks that you use to keep MS’s lunacy out of your hair? Of course you don’t so they continue to run rough shod over our systems.
Remember we are only borrowing their software but the work we create is ours, not thereis yet how many people have lost work or even jobs because of their update policy.
I’m so glad that I said NO to windows 10.
I’m sorry, but the “average user” has proven, time and time again, that they value convenience over security, and if given the option, will delay their updates for weeks or months (or years).
They are exactly the people who caused the forced update policy to start with– and most of them would still be running Windows XP if they could.
Great. So because some people are morons we all must be treated like morons.
Instead of “AI” stupidity, Microsoft should really allow you to defer an update for a certain grace period. Here it could remind you of a final must install date. Like let’s say a week.
They did somewhat add this feature after a while in Windows 10, but the implementation is still flawed because its default action is to force install the update if I for some reason is AFK or need to restart my computer. ‘After Hours’ allows me to somewhat work around the system by telling it I work almost 24/7, but that’s still a hack.
Also I’d like to point out there are many other possible reasons than “you didn’t click save” for why my computer isn’t allowed to reboot. It could be running some process that cannot be interrupted without causing significant loss of time.
(A lighting storm or power outage could cause that too, but they haven’t for over a decade while Windows Update has done it to me several times already just the last couple of years)
Windows update is another example of a piece of software “designed by a committee”, overly complicated and prone to failure.
Updating any piece of software is a two steps process, download and install; windows takes 5 to 6 steps including “preparing to download”, “preparing to configure” … And any of those steps can fail with obscure hex decimal codes not even microsoft itself knows exactly what they mean.
If you are stuck with a failed update you have to look on the internet for a solution and they are mostly generic solutions like deleting the cache or the db. The tool provided by microsoft itself to fix its update process (yes, they need a custom tool to fix the update process, awesome) simply automates the cleaning procedures, but sometime it just does not fine anything wrong. Last option is to manually download the update and install it.
Moreover, my company uses SCCM to deploy update and new software: every update has to pass the IT validation process and windows update points to an internal repository. Then why is there a link to check for updates on microsoft own servers ? After selling the SCCM tool to central manage a fleet of pc’s, is Microsoft inviting me to by-pass my IT ? non-sense.
Using AI to find out when applying a Windows update is less likely to bother you reminds me of clippy: “it looks like you’d like to update windows now”.
The main issue with updates on Windows (as opposed to Unix like OSes) is Windows’s model of file locking.
In Linux, you can replace a file that’s in use with an atomic rename. Processes that had the replaced file open will just keep a handle to the old inode (now not linked to the file name anymore), that will go away as the last handle is closed. Meanwhile newly started processes will see and use the updated version.
This extends, crucially, to shared libraries and other shared data that make much of the OS. That way Linux updates just apply at any time, and programs get the fixes as they get restarted. Server processes are transparently restarted by the update system, and mostly get their state back on restart. The user has no need to complain.
Shared state in libraries is typically implemented via IPC primitives that don’t depend on the specific version of the library bring loaded.
The only exception are kernel updates, which require a reboot (except for kernel patching hacks) because kernel state has no way to be moved to a new upgraded kernel (getting that done would be a huge undertaking with a huge maintenance burden that I’m sure kernel devs wouldn’t consider worth the effort). And even in this case, you can still keep running the old kernel until *you* choose to reboot.
On the other hand, Windows chose to lock “files in use” when its filesystem was designed. An open file can’t be replaced. Instead you get an error saying that the file is in use, and you need to chase the process using it and kill it first to be able to do so.
As this extends to shared libraries and data, it means that all users of a library need to be shut down before an update can happen.
Thus, when a core Windows library needs to be replaced, the whole system needs to be stopped, and an alternate execution environment that doesn’t depend on the updated code has to run to replace the actual files. Hence the reboots and general unavailability during the updates, and the bad experience.
As the update environment is likely a basic and reduced (and maybe safer) version of Windows, it probably also means that the file operations it performs aren’t as well optimized as the ones in the running system, thus contributing to the problem.
While mechanisms like side-by-side assembly (designed to tackle DLL hell) could somewhat help with updates too, apparently its use is problematic when DLLs are used to share data across processes in Windows.
At this point, it’s highly unlikely Microsoft will change file semantics in Windows to allow the replacement of open files (with versioning), so I don’t expect any significant improvement in how Windows manages updates. It’s all going to be a measure of how the update surface (size of the OS likely to require updates) evolves in relationship to the power of common hardware.
This works on Windows too, because you can rename running executables.
The difference between the two is that on Linux a delete will remove the name from the namespace but leave open descriptors to the old inode; on Windows the name remains in the namespace until the last handle is closed. This means on Linux you could delete a file then write a new one while executing the old one, which doesn’t work on Windows. But with atomic rename, this point is moot.
Note that if updating a running program was truly impossible, Windows Update wouldn’t be able to update itself. Pushing the problem to some other lower level component doesn’t solve it either (eg. smss can’t update itself.)
Sure it could… you just need a two-stage process, using a stub that has no function other than to be started by Update, copy over the no-longer-running Update binaries, then restart Update.
It’s not quite that simple, because all of the dependencies also need servicing. If the two programs are both user mode Windows programs, they’ll both use kernel32.dll and ntdll.dll, which neither can update. And then there’s the kernel, and drivers, etc.
Your point is still valid in that it would be possible to have two totally separate OS installations where one running OS updates an OS that is not running, and vice versa. This approach is used on popular phone platforms. But going back to where this started, doing so would increase the number of reboots, not eliminate them.
This is how I always understood it to work, but I have also always wondered how that worked on Linux if:
old prog.exe uses a shared.dll that it hasn’t loaded yet. Let’s assume word.exe and spellchecker.dll.
Now the old word.exe (already in memory) was expecting old spellchecker.dll (not in memory yet). However an update is ran and old spellchecker gets replaced with new spellchecker.dll. When my old word.exe (still in memory) eventually calls spellchecker.dll it will be the new spellchecker.dll which might have different signatures for its functions likely causing the old word.exe to crash.
Of course this is a simple example and you could say “whenever word.exe calls spellchecker.dll it should check if it is a correct/expected version” but now imagine the same thing happing with “shell.exe” and “shellsubdriver.sys”. Normally the package manager will make sure that this dependency is correct, but during updatetime it will get broken for running programs.
How is this handled? Will these crashes happen as I would expect, are programs written “perfectly defensively checking dependencies”, are programs “virtually locking” dependencies?
The linux update process has indeed two intrinsic flaws:
1) the updated library is loaded only by new processes, running processes still use the old library and hence they are not updated. only core services are restarted by the update process, user apps are left as they are, leaving the user under the false impression of being updated with latest security patch.
2) having a mix of process using old libraries and processes using updated libraries can lead to crashes and/or corrupted UI needing at least a logout/login to fix all problems.
This isn’t actually an issue very often in my experience. The only cases where I’ve seen issues like this are really old software that makes some not entirely accurate assumptions about how libraries get loaded (and X11 userspace drivers, but those are always accompanied by kernel modules and thus you should be restarting on such updates anyway).
Also, note that neither point is an issue on some systems. See for example how NixOS or OpenSolaris handle updates. Both create a new system image that you boot into to finish the update, but require exactly zero downtime other than what you would normally have for restarting the system (and also, notably, give you easy to use atomic updates that are trivial to roll-back). Even on distributions like Gentoo, which normally do all their updates into the live system and don’t inherently restart services as they get updated, it’s possible to do things like this.
Interestingly enough, Windows could do things like that too (that is, atomic updates without needing to have the system offline for an extended period of time), they just don’t for some reason.
I have no objection for rebooting my machine after an update, if i’m kindly asked about it and i can defer it.
Linux does not even tell you to restart you machine, once the deb packages are updated, it’s done. At work i reboot my linux machine once a week after updating packages to latest versions, because i know it is better to start fresh after updates. But if an update procedure requires a restart to be 100% effective, why am i not asked to do it ? You are not better than windows just by skipping a tedious reboot request and hoping the user will do it on its own.
In my experience i have never seen a windows pc rebooting on it own to install update while the user is actively working on it (unless IT is actually mandating to install the update and you have already deferred it for a month). On the other hand i have seen linux machine with one year worth of updates pending to be installed because linux users cannot be bothered with such trivial issues.
Edited 2018-07-26 13:07 UTC
This has nothing to do with the primary arguments in favor of how updates typically work on UNIX-like systems (not just Linux).
The two primary advantages of the way UNIX-like systems (including Linux) handle updates as compared to Window are:
1. You don’t need to reboot for self-contained updates. As an example, if you update IIS on WIndows, you need to reboot, period, even though IIS isn’t a core part of the system. In contrast, updating Apache, or nginx, or any other web server on Linux only requires you to restart the web server after the update, not the whole system.
2. When you do need to reboot, the reboot will take no longer than it normally would (unless you’re doing a full system update to a new version of Fedora or RHEL, which works more like a Windows update).
A really good example that illustrates both benefits is the case of a small bug in a widely used system library that only affects a single program that uses that library. On Windows, you have to restart the entire system, and it will do 90% of the work of updating the library while you can’t use the computer. On Linux, you just update the library and restart the one application that was having issues, and if you do need to reboot the whole system for some reason, it won’t take any longer than it normally would.
Put a bit differently, imagine being able to actually use your system (even at a basic level) during that half hour it takes to update to a new build of Windows, with the only downtime being the minute or two it takes to reboot when the update is finally done.
On my 4 year old laptops a new build (the ones you get every 6 months) takes 10-20 minutes of downtime and 30-20 minutes of online updating. The regular patches that you receive 1-2 times per month only take 2 online minutes and 1 offline minute
http://www.osnews.com/thread?654765
http://www.osnews.com/thread?654770
Good to hear that it isn’t a problem that happens a lot in real life. I would like to know more about the technical does-and-don’ts to prevent such issues
Any system that doesn’t automatically update: BAD
Any system that causes me to be loose control over my resources/time: BAD
Any system that silently updates applications that are still in use causing unreliability: REALLY BAD.
Any system that restarts while there is unsaved work: REALLY REALLY BAD.
I don’t know any non-BAD update system, but since Windows has at least 2 of the issues above it really is the worst. And AI that guesses when I am least like to mind loosing control over my resources/time isn’t going to change that verdict at all
If every now and then my motorbike would take control during a ride, and veer off to park on the curb for some self-maintainance, maybe every week or so, when I was least expecting it, then I wouldn’t ride it.
That wouldn’t change much if someone suggested I could set Active Hours of Riding or other fine tuning. Like I know that. Like I should have to know that.
Nor would it change much if it was patiently explained to me that it was statistically safer to automatically veer off to the curb for self-maintainance than to let me ride with a slightly mis-aligned back wheel, or clogged filter.
It would make no difference whatsoever if — after complaining to the manufacturer about this absurd behaviour, which made the vehicle unsuitable for serious use — they said, “We hear you, so now your bike will try to intelligently predict when it is OK to veer off to the curb for self-maintainance.”
Instead I would be tempted to look into bikes that were better conceived and constructed, and therefore didn’t need to indulge in that crap.
What is this, a motorbike analogy? Don’t you know that car analogies are an established standard?…
Well, where to start?
* Obvious complaints from other people here about lost work (which I don’t sympathize with, save your damn documents when you walk away from your computer, this isn’t the only thing that can cause you to lose work overnight if they’re not saved).
* Updates don’t work reliably on some systems. This is usually a problem with the third-party drivers for that particular system and thus not really Microsoft’s fault, but they could at least handle things a bit more sanely.
* The whole restart to install updates thing is crap. Yes, you should restart to actually use them (well, at least restart the things that were updated, there’s no reason you should have to restart your whole system to finish an update to a web server), but requiring a restart to actually install the files is just them being stupid (yes, I know that you can’t replace, rename, or delete open files in Windows, but that does not preclude you from using things like the snapshotting infrastructure they already have to do updates without needing to disrupt running code).
* If your system isn’t on 24/7, updates run some random amount of time after you start the system, which is really irritating for people who want to be able to use the system without interruption immediately (gamers who only use Windows for gaming for example).
It sounds like the AI might help with the first case, and possibly the last one (if they are going to include the download as part of the update process it’s controlling). However, the second issue needs to be addressed by the OEM’s who are building shit drivers, and the third can’t be fixed at this point.
Are they going to fix the issue where laptops/tablets come out of standby mode at 2AM to install updates and then never go back into standby mode until the battery is drained?
Microsoft’s decision to force its updates on users without allowing them to review updates or opt out of certain updates was a mistake. I’ve heard many horror stories of stuck computers, computers that won’t start properly after updates and hardware that no longer worked after updates. People have also lost data or unsaved work due to forced updates. I understand why Microsoft wants to ensure everyone has the latest patches because Windows is so prone to malware and viral attacks. But that’s on Microsoft. People tend to forget that Windows is still based on ancient code dating back to the early 1990s. With so many band-aids plastered on the code over the years it’s no wonder Windows is so insecure. Microsoft needs to do what Apple did in the late 90s and just start with a brand new modern operating system.
But that’s what they did, with NT… it’s quite secure for some time now. Heck, current macOS (NextStep, really) is slightly older than NT (and did you already forget major security missteps of macOS from the last year, of the severity that Windows last time had in the 9x days …but I suppose you don’t forget those so easily, right?)
Linux etc do this sooooo much better and soooo much faster.
Probably this has all been said already in the comments thread, but current Windows Update aggravates me so much that I can not let the chance of venting off a bit go wasted.
a) whenever windows tries to apply updates overnight on its own, my laptop crashes. It has never managed to reboot itself properly once automatically – although it does when I apply manually the updates.
Yes, I never ever shut down, but do put laptop to standby every night and yes, I keep multiple VMs open at all times, plus the mounted encrypted drives and ssh agent. I don’t like to restore all this ‘working condition’ state after a crash.
No, my company does not run enough windows laptops to warrant a ‘win enterprise’ license.
Dell XPS 1560 on latest win10 64bit for the curious.
b) why on earth do I get the laptop attempting to install driver updates and even BIOS updates behind my back? Why not allow only ‘application’ updates go though unattended and require my approval for the risky stuff?
c) the “simplified” list of installed updates is basically a huge lie, making it worse than useless. The items listed do not correspond to what I see the laptop installing in front of my eyes. Also, I can not have a single list of all installed updates sorted by date, I have to stick to checking for updates in 5 or more meaningless ‘categories’
d) trying to guess when I am not using the laptop is hard, as I use it both for work and personal communication / media streaming / everything else. Using an AI will not help that much as my patterns are as random as you can get
e) I can not postpone installation of updates to night time anyway, as a freaking VPN client that I had to install for a customer insists, before it lets me in, that I install every single update *as soon as it is out*. Which means that, every time an update for windows defender is released, I have to still apply it manually before starting the vpn connection.
f) nitpick 1: why does the login background image change when the laptop reboots after an update? Is that a contrived way to signal to me that something has happened?
g) nitpick 2: the ‘window update’ panel from the system-settings has buggy display behaviour in corner cases which involve scrollbars flickering because of the length of text labels
In short: the whole point of ‘apply overnight’ / ‘apply unattended’ / ‘take away control’ is lost on me. And the whole thing is s stinking pile of bugs. Fix them first, and leave AI second.
One: I don’t like the 5 restarts Windows 10 needs when updating. It dualboots in Linux so I have to pay attention.
Two: experimented with downloading from other computer settings and it woke up in the middle of the night (1 am) just running linux because that is the default. Disabed it and never seen it awake again…
Anyone who thinks that the Windows 10 update process is ok, has obviously never had to release updates in a corporate environment.
Windows Update has never been great, but its behaviour in Windows 10 seems to have worsened. You should always have the option of manual updates so that you can decide when to apply them, but that seems to be gone in Win 10, even in the Enterprise Edition (yes, you can delay them, but it *will* eventually auto-apply them).
Don’t get me started on the fact that many updates seem to require a reboot (and can’t apply all of the update before the reboot – always reboots at 30% for me on Windows 10) or that non-Microsoft software can’t use Windows Update (so there’s a trillion ways such software can update – and quite often it’s a long trawl through them all manually to see which have updates).
Basically, it’s Linux Updates 1 Windows Updates 0…
Linux updates have their own pain points, mainly the fact they cannot gracefully upgrade configuration files that have been modified, not always anyway. Maybe we need a KV value store instead of storing configs in text files, a “registry” of some sorts.
BTW what I truly hate about Windows 10 is that the humongous Redstone updates (upgrades actually) -which take 2 hours just to do the reboot routine- are being pushed on a non-optional basis. This is the result of Microsoft pushing the “OS as a service” model to everyone who cannot run away fast enough. This model basically dictates that everybody has to use the latest and greatest version and everything else is unsupported patch-wise. I mean, not even Google tried to do this. And even Canonical has LTS releases, although sometimes app support is slipping on those sometimes. It’s the same reason “rolling release” distros are reviled by the general populace. I moved my Windows 10 computer back to 8.1, never looked back. And I have a usable control panel now too.
Edited 2018-07-27 19:41 UTC
I don’t mind automatic installation of updates – the problem, as I see it, are the forced, automatic reboots that come along with them. Part of it is the fact that many Windows updates even need a reboot in the FIRST place, but that’s another matter.
I understand why they do it: when you have updates that require a reboot, installing them automatically isn’t much use if the users don’t restart to complete the updates. But I suspect there’s a large overlap between users who would put off reboots to install updates & those who aren’t very diligent about saving in-progress documents, and therefore most likely to lose work as a result of forced automatic reboots.
And with people who have more complex multi-document/multi-application workflows, there are less-tangible things that can’t easily be saved (especially since Windows is still largely incapable of restoring closed applications/documents after a restart, with exception of Explorer windows). Not to mention things that can’t really/easily be saved for technical reasons, for example data entered into a web-based form that hasn’t been submitted yet. If your browser doesn’t happen to retain that (and even the ones that do are hit-or-miss in my experience) & Windows forces a reboot to install updates before you’re finished, then you’re SOL.
The way Windows updates has worked since Win8 seems like a textbook example of a cure worse than the disease: to address the issue of incomplete updates, they’ve created an (arguably/potentially) much worse problem, mostly thanks to ignoring the most basic, fundamental principles of data-loss prevention. It’s the digital equivalent of having cleaning staff that comes in overnight and, if your desk is cluttered with in-progress work, they “fix” that by just sweeping everything into the trash.
Personally what I would like is the option of automatically downloading updates and having them install when I restart, boot or shutdown the machine.
The active hours option seems an acceptable default if you can switch to something else like the above if you want.
I am not suggesting that updates should be disabled, but having the machine restart itself is a tad frustrating when I want to leave something running.
Using a ‘trained’ ‘predictive model’ seem even less predictable to me.
Microsoft is either tone deaf or doesn’t care. At this point I don’t think they could find their own arse if they used both hands to look for it.
Is this a joke? MS concerned about upgrades? After forcefeeding users? Ridiculous. What next, linendings that work in notepad (oh wait, those are coming in 2019), I guess pigs can fly!
Is it only Americans who keep their PC devices on overnight?
I in the UK find it bizarre that Microsoft’s default schedules for updates and scans are performed after 2am in the morning.
Most of us in the UK, turn off our devices when we’re finished with them.
I’ve went to countless people’s homes doing Computer repairs and in every single case, they have never altered the early morning settings for Windows.
No wonder their computers never get a daily AV scan or updates when they’re not using it.
What gives Microsoft?
Thom is from the Netherlands. And I, in Poland, also typically leave PC on… (largely because there’s almost always some music playing)