Only reason I can see
Is power saving on OLED monitors. Other than that, what's the point of this other than hoping to make anachronistic all the jokes about Windows and "blue" screen of death?
The infamous Windows Blue Screen of Death (BSOD) will be replaced later this summer by a new black screen as part of Microsoft's Windows Resiliency Initiative (WRI). Initially previewed in green, Microsoft's bet on black at least ensures backward compatibility for the BSOD acronym. Images of the revised color scheme can be …
It works for the Haggunennon Admiral's space ship:
ZAPHOD: It's the wild colour scheme that freaks me. I mean, when you try and operate one of these weird black controls which are labelled in black on a black background, a small black light lights up black to tell you you've done it. What is this? Some kind of intergalactic hyperhearse?My advice is to run away. . . terribly fast.
I expect it's simply because the Windows boot screen is black. At the point you see the message your machine has crashed, set some flag or EFI entry so the bootloader knows this and rebooted.
The booting screen then shows this message, like when you've done a reboot after an update... (The "You're 33% there.... You're 73% there... You're 100% there... You're 100% there... You're 100% there... <for literally 5 minutes once>" screen)
Because "end of the blue screen of death" is a better "story" for dull and/or lazy mainstream media types who have an excuse to use it as a lazy hook for discussion or contrived point of interest with shite like "now, we've all had our computers crash with the 'Blue Screen of Death' at one time or other, but Microsoft has just announced they're killing it off". Even if nothing is changing beyond the colour. And MS knows that.
For example, there was an iterm on this on Radio 4 when I was walking to work on Friday, but it's safe to assume there wouldn't have been otherwise.
This gets MS onto the radio, TV, etc., gives the impression that they're doing something and provides an excuse to mention all the other alleged improvements like faster restart after a crash.
Who'd have thought that simply changing the colour would help with all that? *cough*
FTFA: Redmond has been here before: Windows 3.1 had a black screen of death and a "blue screen of unhappiness."
I seem to recall that by editing a .INI file in Windows 3.1, the "screen of unhappiness" could be whatever colour you wanted, within the confines of the CGA palette.
Similar pedigree and only ever heard "bee-sod", literally. Non-technical people just said (and say) "crashed" or "broke" and anyone who knew of the acronym has always pronounced it as that, in my experience. Literally never occurred to me anyone would spell it out.
Simple : use a Linux kernel.
Oh, and teach your developers to work in an environment that demands they actually pay attention to security.
Then you can go on make Excel charts and Powerpoint presentations that manglement can't get enough of in an actual computing environment.
I'm sure your bottom line will manage . . .
"Simple : use a Linux kernel."
Linux still has a few situations where it drops into 'maintenance mode' when it fails to boot back up. Particularly when a mounted drive has failed. So BOTH are in need of a little more work to allow them to work around problems that can be worked around.
But I'll not be sending the W10 machine to the recycling site any time soon, so any improvements to W11 are useless to me, but the drive to kill off X11 on Linux is just as annoying as telling me that my computer can't run W11. PLEASE can we have something that simply works as it has since W3.1 days ...
No, install a broken driver in Linux and it does crash as well - again people who believe Linux is "secure by default" don't understand anything about OS or security.
The real problem is the truly outdate two rings model used still today - and the four rings model was dropped exactly because old OS designs - like Linux - can't use it.
If the hardware could segregate the core kernel from other system processes that have higher privileges than user code, but not high as the core kernel, it would be possibie to build far more secure OS.
It looks MS is attempting something alike, but without hardware support.
As long as we keep on relying on OS designed fifty years ago we're not going anywhere, we just pile up workarouds over workarounds.
"The real problem is the truly outdate two rings model used still today - and the four rings model was dropped exactly because old OS designs - like Linux - can't use it. ... - it would be possibie to build far more secure OS."
Pining for Multics?
Only half joking.
Multics had some interesting ideas, others not so much. Of course any previous ideas influence the subsequent ones. Anyway, security can't be only achieved at the software level - hardware support is necessary. Chasing performance only led to non secure designs, or no longer secure ones.
IMHO there were a lot of good ideas in the 80286 protected mode design - and even more so in the 80386 when segments were no longer limited to 64K and pagination made easier managing memory.
But OS designers just used two rings, and used flat memory spaces because it was easier, faster, and it was what OS already used. And decades later, we are still fighting against the same security issues, a lot of which would not be possibile in better, "newer" hardware/software designs. Just nobody is willingly to invest in a new, more modern and secure OS. Especially now that money is made not selling hardware/software, but hoarding as much data as possible, and the less is invested in R&D (but the hoarding itself), the higher the profits.
While people have been brainwashed to think that a 50 yeears old design, and not the best one, is perfect and doesn't need changes. The "Cult of Unix" is alike Scientology (ensure those at the top get richer and richer, exploiting those below), amd just started the IT "dark ages". Maybe one day will see a "renaissance" - but I can't see anyone investing the needed resources soon.
So let's change the kernel error screen color.... that's real innovation!
I'm rather partial to the flat memory model. Near, far, and huge pointers weren't fun.
The various PDP models' memory was segmented. Flat memory came along later with m68k and the 386. Over time CPU design evolved and allowed areas of memory to be marked with different properties (e.g. no-execute) and as a consequence OS design evolved to use these features - it hasn't remained unchanged over the past 50 years.
As for rings, what extra thing or two things would you partition? Two of the four rings that VMS used had nothing to do with security and they never made it to NT.
The kind of security we need today is more about finer grained process isolation from each other with limited interaction between certain processes. As in Qubes OS.
Even if someone wanted to write a new OS which used rings 1 and 2, Intel's probably going to get rid of them soon.
I didn't downvote your post because you're correct that Unix wasn't designed to be a secure OS. There's an old Bell Labs paper from Ken Thompson that explains that. My recollection is that it says that Unix was designed to discourage users from inadvertently walking on each others files while still allowing some amount of collaboration. I read that paper three or four decades ago and maybe I misunderstood. But if I got it right, Unix actually seems to do that quite well. I did try to find a copy of the paper on line to reread and link to. But Igor or whatever Google calls their AI agent didn't come through with anything.
As for design. My impression is that we actually have little or no idea how to design a secure OS that is usable. I doubt we have a clue how many, if any, "rings" that secure OS would have. I'm not sure we even know for sure whether a secure OS that allows users to get work done is possible
Of course there far less concerns for security back then, and the hardware was what it was available back then. But today it's pretty clear that we need something more to have secure system. Of course a system still have to let people use and share data - but there's a lot that could be done to avoid issues like Crowdstrike.
There has been a lot of researches, and attempts like the Intel one, but nothing came to production. In the past I can understand that peformance improvements were paramount to sell more hardware and software, but today not so much. The fact that MS had to kill still adequate PCs witn Win11 shows how much desperate they are to force people to upgrade.
We got workarounds - VMs, containers - why a modern CPU and OS can't really partition software and applications at the hardware level for better security, when needed? You might wish to run some applications in one partition for easy sharing, others in different partitions for better isolation and security. But still, third party drivers run together the core kernel.... and again, when new CPUs are shown everything they talk about is performance. While data hoarders need more and more perfomance to extract the last drop of users' blood, most other users doesn't really need it, and would benefit more from more secure architectures.
PS: I do not know if The Register has an automatic downvote feature for readers who are critic of their editorial line, or if there are users who spend all the day to track and downvote post from people they don't like, LOL! But really, who cares about votes!
Allowing any Microsoft software to carry on after it has signalled a problem is something that only beginners do. I'd give it a week, max, before they too learn the uncomfortable truth.
That problem alert is a signal to you as user that a timebomb has been set, and unlike in Hollywood films there is nobody able to stop that clock once it has started, not even last minute.
It means rescue what you can and then stand well back. Best use a long stick to make it reboot.
I get at least 3 a day. WinDbg says that they are all caused by an unsigned 64 bit integer. Buggy network drivers apparently causing several errors in netio.sys. Doesn't inspire confidence in Microsoft products. In some cases it's like 1999 never ended with the Internet Explorer 4 scroll bug; opening Edge BSOD's.
I used for years Dell computers at work and saw a BSOD once because of a buggy F5 VPN driver. A couple of years ago they have been replaced by HP models to save money, and it became a BSOD festival. Being the OS and software the same, I would say the problem is the hardware. Updating many firmwares and drivers with time reduced the problem.
What I don't understand is why all the "telemetry" they capture doesn't help to fix this bug far quicker.
Wait, wait, let me get this straight, you're assuming that telemetry is there for your benefit or that of customers?
Oh. my. God.. Hahaha, hihi, mmmmpf, mmmffBWAHAHAHAA hahaha, sorry, hihi, give me, bwahaha, hihi, a moment, hahahaha. Thanks, hahaha, sorry, that was unexpected. Hihihi.
You must be new here :).
In addition to the above post, some BSOD's leave 0 byte DMP files. WnDbg can't open them. Other software just shows a (blue but) blank screen, along with a short report of 0 crashes. I've had 6 of these recently. Maybe Mickeysoft is trying to fix this. During my use of Windows 10, I've had numerous Power-Kernel events.
One of the approaches to making Windows more resilient involves revising how security applications (SA) interoperate with the Windows kernel (WK). As part of the WRI and the related Microsoft Virus Initiative (MVI), Windows is being tweaked to run security software outside of the kernel (SSOC). That should make it less likely for the entire system to fail due to shoddy vendor security code. (SHIT)
FTFY.