Linux 5.0 is out except it's really 4.21 because Linus 'ran out of fingers and toes' to count on Linus Torvalds has squeezed out version 5.0 of the Linux kernel and flung open the merge window for its follow-up, 5.1. In the post announcing the arrival, Torvalds was at pains to point out that feature-based releases really aren't a thing and the 5.0 "doesn't mean anything more than that the 4.x numbers started getting big …
I doubt it.
Ubuntu 19.04 is due out April 18th, so will almost certainly be released with kernel 5.0.
Linus won't even issue a 5.1rc1 until around March 18th, and usually goes to at least rc6 before declaring a kernel fully baked.
By my mathings, that won't be until April 22nd at the earliest.
"Well, as Windows fans will be well aware, 5.0 is ... the moment at which Microsoft abandoned such numbering for its server products. There is thankfully no sign of Torvalds adopting a similar approach with Linux."
I should think not. In the Unix world we're made of sterner stuff. 5.0 is quite a modest version number.
Installed yesterday:
Information for package kernel-default:
---------------------------------------
Repository : Haupt-Repository (OSS)
Name : kernel-default
Version : 5.0.1-1.2
Arch : x86_64
Vendor : openSUSE
Installed Size : 346.7 MiB
Installed : Yes
Status : up-to-date
Source package : kernel-default-5.0.1-1.2.nosrc
Summary : The Standard Kernel
Description :
The standard kernel for both uniprocessor and multiprocessor systems.
Source Timestamp: 2019-03-11 06:36:03 +0000
GIT Revision: 8c6a826d31764166517a04ee7f27ad52a71a6855
GIT Branch: stable
...of the programme numbering of the Burkiss Way (with Burkiss spelt wrong) where they not only started a new series half way through one programme (which is referred to as episodes 31 and 32) but also had two episodes called "Repeat Yourself The Burkiss Way" one broadcast a week after the other where the first few minutes of each were identical and are both referred to as episode 39.
For anyone not familiar, it was a sketch show that was the nearest equivalent to Monty Python on radio and is still often repeated on Radio 4 Extra. See:
https://en.wikipedia.org/wiki/The_Burkiss_Way
IIRC, the 5.0 meant for Windows 2000 was NT 5.0; at that point, NT became Windows and the old DOS-based Windows line (3.1, 95, 98) became extinct after Windows ME, when Windows XP with the NT 5.1 kernel came out. NT was a modern kernel, whereas DOS-based Windows did not make full use of the 386 architecture.
DOS started to exploit the 386 when EMM386.EXE was introduced. DOS programs were able to run in protected mode using DOS extenders (they required to be specifically written to use them), and some supported 386 features.
Windows started to support some 386 features with version 2.1 (called Windows/386), not to allow for 32 bit applications, but to take advantage of those 386 features that made supporting some features easier - i.e. the Virtual86 mode, being able to switch to/from protected mode without a CPU reset, and improve memory management.
When Win32s was released Win 3.1 got some support to run 32 bit applications.
With Windows 95, the APIs and applications became fully 32 bit, even if the OS itself used some older 16 bit code for some tasks.
You are right, I had forgotten about those DOS extenders, possibly because they were often a real pain to get working OK.
I think some worked with the 286 as it could use up to 16MB of memory (but as you pointed out with switching to protected mode and back which was a big pain until the 386, as the 286 designers never expected you would want to switch *back* to 8086 "real" mode so you needed to get they keyboard controller to generate an interrupt to pull it out of a halt and back on the real mode).
Windows 7 was and still is actually 6.1, AKA Vista+ despite it being, IMHO, the best OS Redmond ever produced, as evidenced by it reaching end of life with just 1½ service packs and them having to actively sabotage it with telemetry to get people off of it. Just shows you how much shit they have in the upstairs cupboard when it comes to version numbering, eh?
Windows 2000 was a good release as well - as it removed many of the limitations NT4 had compared to 95, as it supported more recent hardware - i.e. USB, laptop power management, DirectX, etc. Many kernel changes came because of that.
With far less systems connected directly to the internet back then, security vulnerabilities were harder to exploit.
XP had the issue that it didn't support 64 bit systems - XP 64 was a half-backed solution, unlike 2003.
Windows 2000 was a good release as well - as it removed many of the limitations NT4 had compared to 95, as it supported more recent hardware - i.e. USB, laptop power management, DirectX, etc. Many kernel changes came because of that.
That made me want to upvote!
XP had the issue that it didn't support 64 bit systems - XP 64 was a half-backed solution, unlike 2003.
That made me want to downvote.
XP was Windows 2000 + USB2 patch and a resource-hungry toddler ui, for the simple minded audience Windows has always targeted.
XP64 and w2k3 64 were PITA's, driver support, anyone ???
Compare to Linux ....
WoW64 is a mess, 64-bit binaries in %windir%\system32 and 32-bit binaries in %windir%\syswow64, whoever came up with that idea should have been laughed out of the room.
I used 2003 64 bit extensively - as our databases and applications no longer fit inside less than 4GB of memory.
Good servers did have drivers for 2003 - we were using Dell and HP and never had problem with divers, as 2003 64 was fully supported and drivers made available.
If you tried to install it on your desktop, laptop or DIY server, and hope to find drivers for consumer hardware, you would have been probably out of luck, as you were with XP 64.
Anyway, Linux driver support is still one of its major issues too - especially for hardware that isn't mainstream.
"XP was Windows 2000 + USB2 patch"
No, it had far more changes, in the kernel too. It's always funny how much technical people just look at the UI - GDI+ was not the only new feature - and never know what changed under the hood. For example, XP introduced substantial memory manager changes as the RAM installed on PCs was rapidly increasing. IIRC, it was also the first MS OS to use the then new faster CPU instruction to call into kernel code.
"64-bit binaries in %windir%\system32"
That was again a move needed to ensure backward compatibility - even for badly code applications and installers, which Linux ensures only forcing users to keep on using very old and outdated code applying only minimal changes like RedHat does, otherwise mostly everything breaks.
Not elegant, true.
AFAIK Win2k NEVER supported someing as base, and plebian as DirectX. As Win2k was sold as a being the sophisticated Busnes orinated NT Client, to the more plebian class that wanted DirectX and, were seved Windows Me instead.
Inb4: Ahh yes but, you could bodge DirectX onto a clean non DirectX copy of Win2k. I'm not even gonna argue the point. Yeah it could have been done, and I'm sure enough People did, back then. But, I think it only fair to point out that this was... Say unlike with the later WinXP, an unsuported setup.
I mean you may as well say that MicroSoft sabotaged Win98SE (Or the whole of 9x, anf thus the reason for WindowsX), was eventually sabotaged by WindowsMe. Which had hardblocks in place to make you ditch your old DOS Software. Even though, being a 9x Kernel, and so fully able to perfectly run any DOS Software. You had to dig in deep to the registry service(s), to undo the vodo that, they done so well. That it only made sence to dual boot 98SE alongside a beta release of XP, and understand that DOS 's time was (as they say...), up.
Just because MS decided SP were no longer needed as updating through Windows Update/WSUS could have been enough. Some years before, 7 would have got more SPs too.
Now, with Windows 10,m they don't release SPs, they just install a new copy of the OS.... it requires less effort and testing than upgrading an existing installation.
Granted, but I liked the implication. Agree with the Win2k comment upthread, too. I doubt 10.1 will be any better, though. The bits that make 10 brown and sticky are core features now, much like XP's Fisher-Price UI and bloody activation cock-ups.
Sully themselves with nVidias Drivers? Ok could someone please explain why I would want to use the Nouveau over the ones by nVidia? Other than the pure joy there is to be possibly found in being fundamentalist open sauce zealot*?
I mean is this like the only difference? How exaclty do the Nouveau Drivers hold up compaired to those from big, bad evil nVidia?
*Not that there's nothing wrong with this of course, but some of us aren't as strong with the faith as other.
Personally I've had way too many problems with the nouveau drivers to even consider using them. I've still got one laptop that I need to sort out after an upgrade where everything points to this driver. I just can't get the desktop to start. Really need to find the time to do a backup and clean install.
I'm having no issues with the i965 drivers with Mesa on my fallback laptop.
I use the NVIDIA drivers on Linux when I need to do GPU debugging and performance analysis.
The following is a great site to see the state of Mesa support for OpenGL version and the different drivers:
https://mesamatrix.net/
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
