Windows end-of-support scaremongering

[peter-h]

Here is another one...

The other day I tried to install Python 3.14 or later and pycurl, on a little win7-64 fanless computer. It was to run an automated task.

Now you would think that Python, being basically a script language interpreter, would just need some OS with the usual API for files, comms, etc. No, the installer requires win10! Why??

[paulca]

QEmu, Proxmox, Oracle VBox there are many, many options.

Do these have a simple and free Windows player like VMWARE has?

I had a go with Virtualbox a few years ago (I used to know a guy who wrote some of the original code, before it got sold to Oracle) and found it very difficult to configure properly.

Windows you can use the built in Hypervisor Windows itself runs on.

Just run HyperV manager.

[paulca]

No, the installer requires win10! Why??

Do you really think that people releasing python are going to check and test it on Windows 7?  What about on XP?  3.1?

[thikone]

Here is another one...

The other day I tried to install Python 3.14 or later and pycurl, on a little win7-64 fanless computer. It was to run an automated task.

Now you would think that Python, being basically a script language interpreter, would just need some OS with the usual API for files, comms, etc. No, the installer requires win10! Why??

There is a way to run it on Win7. Some DLL needs to be replaced. I think it was also the reason why KiCad 6 demanded Win10.

[peter-h]

Do you really think that people releasing python are going to check and test it on Windows 7?  What about on XP?  3.1?

Well, Python did not come into existence when win10 came out

So at some point somebody decided to make the installer check for win10, just for fun. Probably because they watched too many YT videos

FWIW I will try to solve this by installing Debian 13 on the machine natively. Google says it should work (Atom 2550, 4GB RAM). Hopefully I will find out if it doesn't like the CPU, before it reformats the win7 SSD which took me several days to set up

[paulca]

https://peps.python.org/pep-0011/

This is the rational.

It's not just Windows.  Note:

Legacy C Locale
Starting with CPython 3.7.0, *nix platforms are expected to provide at least one of C.UTF-8 (full locale), C.utf8 (full locale) or UTF-8 (LC_CTYPE-only locale) as an alternative to the legacy C locale.

Any Unicode-related integration problems that occur only in the legacy C locale and cannot be reproduced in an appropriately configured non-ASCII locale will be closed as “won’t fix”.

While this will not effect the majority of distros, it may very well effect embedded systems, micro-systems etc.  Those OSes you build from linux, but to save on the multi hundred mega-bytes of the full glibc-locale stack, you just add US-ascii.  In this case, you will now need, as a minimum the UTF-8 locale enabled.  That would exclude a lot of embedded linux OS's which may well be running python in core scripts.  Of course... that's only if you don't upgrade python and are happy with the vulnerabilities discovered in your current version.  I mean embedded systems are typically not very secure anyway, a primary vector now in organisations.

[peter-h]

Why mandate win10 though?

That link just says

Windows versions prior to Windows 10 follow Microsoft’s Fixed Lifecycle Policy, with a mainstream support phase for 5 years after release, where the product is generally commercially available, and an additional 5 year extended support phase, where paid support is still available and certain bug fixes are released. Extended Security Updates (ESU) is a paid program available to high-volume enterprise customers as a “last resort” option to receive certain security updates after extended support ends. ESU is considered a distinct phase that follows the expiration of extended support.

Windows 10 and later follow Microsoft’s Modern Lifecycle Policy, which varies per-product, per-version, per-edition and per-channel. Generally, feature updates (1709, 22H2) occur every 6-12 months and are supported for 18-36 months; Server and IoT editions, and LTSC channel releases are supported for 5-10 years, and the latest feature release of a major version (Windows 10, Windows 11) generally receives new updates for at least 10 years following release. Microsoft’s Windows Lifecycle FAQ has more specific and up-to-date guidance.

CPython’s Windows support currently follows Microsoft’s lifecycles. A new feature release X.Y.0 will support all Windows versions whose extended support phase has not yet expired. Subsequent bug fix releases will support the same Windows versions as the original feature release, even if no longer supported by Microsoft. New versions of Windows released while CPython is in maintenance mode may be supported at the discretion of the core team and release manager.

which is 100% a circular totally-BS justification

It's like me saying to a 5 year old kid "NO you cannot have more than five ice creams today because it is my policy!!!"

[paulca]

It's about "support".  Both the python community supporting python itself and which platforms python itself wants to support.

This is probably the primary argument against "the old will still do".  People do not like to support older software.  Developers do not like to write code for legacy systems or fix them.  Developers don't want to or cannot be limited by "backwards compatibility".  The technology moves too fast.  That which you wrote out by hand in 2007 is automated and imperative implicity today.  There is no point even refactoring or "fixing up" code like that, it's rewritten half the size.  Nobody wants to bring baggage forward so they select the most common feature set and support requirements and move towards that.  The outlyers get dealt with later if at all.  The market moves too fast, unless of course you don't actually want to make any money.  Even then the market size for people still using Windows 7 is smaller than those using Linux and only a small subset of software (desktop, etc.) supports Linux even.  Why would you expect people to spend time, effort and money fixing issues with Windows 7 in their new release.

Can you now see how difficult that would be?  If you had to retain backward compatibility with an OS now, what?  20 years old?  That requires you either use none of the new features or you write horrible, buggy, interface layers and compatibility matrix style macro-soup.  You have to have working tests and worknig test platforms, VMs for all the OSes and its versions.  Service pack levels.... etc.  People to run all that and people who understand how to fix that quirky old Windows 7 bug.

The users who show up and say, "I have a slightly pinker shade of pink on my Unicorn System 3.0 which is customised in the following dozen ways.  Yet your software doesn't support my use case and I'm upset.", should get shown the "Community help" forums and no money wasted on them.  The industry basically feels the same way.  You go with the flow, stay on the wave or get left behind.  Rescue and support get thinner and thinner the further you go.

Luckily there are many smaller grass-roots organisations trying to bring retro hardware back by repurposing them for modern day use, but it's a niche market.  If you want to insist on being that far out of normal you might want to talk with one of them for advice. 

[peter-h]

But M$ stopped supporting their products around 30 years ago...

I remember when they did.

It's a fake debate about winX being out of support.

[iMo]

Windows you can use the built in Hypervisor Windows itself runs on.

Just run HyperV manager.

So you enable Hyper-V in your win11 and then:

1. Convert the existing Windows 10 installation → .vhdx  ( with help of https://learn.microsoft.com/en-us/sysinternals/downloads/disk2vhd

2. Create a new VM in Hyper-V Manager → attach that .vhdx

3. Boot → Windows reconfigures itself → done..

Did somebody try that? Any other hints/experience?

[paulca]

Never tried to convert a bare metal instance to a virtual one.  I just installed Windows10 normally and bought a 'grey upgrade license for it' and then another one that stuck a month later.  £4 each.

It actually somewhat disturbs me how many "valid" windows licenses I actually have.  They transfer with the computer of course, so if you buy a second hand PC and connect it to your MS account, it shows up as a valid license.  I must have about 10 licenses.  Although... it should be noted that OEM style licesnses on things like Dell machines cannot be transferred to another machine.  The single full retail license I have can be.

... and yes, before anyone asks, you do need a full valid license for every single Windows VM and you need to activate them, unless you are deploying them from a central control farm and use the automation tools.

Most people exclude this for very ephemeral stuff, like "boot test discs" for building hardware.  Usually you install your test stuff on it and ignore the "Genuine Windows warning" until it shuts you down, then reinstall it.

[peter-h]

you do need a full valid license for every single Windows VM

It depends on whether you answer "copied" or "moved" to the question of what you did with the newly configured VM If you say Moved, it runs just fine.

But you are right about the strict legality.

[thikone]

Here is another one...

The other day I tried to install Python 3.14 or later and pycurl, on a little win7-64 fanless computer. It was to run an automated task.

Now you would think that Python, being basically a script language interpreter, would just need some OS with the usual API for files, comms, etc. No, the installer requires win10! Why??

Here is why: https://www.reddit.com/r/Python/comments/rbhd2n/python_39_310_on_windows_7/
Developers were lazy and used Win10 only functions to implement some path manipulating functions in python.
Clearly, can be done without using those Win10 functions.
And that guy did exactly that.

The latest that he did is 3.10.1 though. Here are the installers:
https://github.com/NulAsh/cpython/releases/tag/v3.10.1win7-1
For 32b Win7: python-3.10.1-win7-202112121259.exe
For 64b Win7: python-3.10.1-win7-amd64-202112121259.exe
And "full" versions you only need if you want to debug python itself.

I did "recently" migrated to Win10. But not just stock one but custom ISO that I made with help of MSMG Toolkit.
It does not have Defender, Updates and MS bloatware like OneDisk and YourPhone.

Now I dread the moment when Steam or Fusion 360 or some other important to me software stops working on Win10.
At work they already gave me Win11 machine that I hate. I can rant for hours about that but the most astonishing:

• I have to wait before typing in my password. Because if I press Ctrl+Alt+Del and start typing immediately, as I used to do, it swallows beginning of my password!
• I have to restart TaskManager every few days because it just stops drawing popup menus and then starts to hang.

And I installed ExplorerPatcher to bring back Win10 style system bar and start menu.
Most importantly, Win10 popup menu for files. So that I can copy file path with one click less, which I do a lot, as a developer.

Fun fact about updates: some Win10 update had a bug in CreateWindow system function. Windows could not create a window, Karl!

[peter-h]

Now I dread the moment when Steam or Fusion 360 or some other important to me software stops working on Win10.

F360 probably won't stop working but the updates have stopped already, which means you cannot open .f3d files from an F360 user who is still getting updates. Bastards!
https://www.eevblog.com/forum/eda/autodesk-fusion-free-version-does-it-do-anything-beyond-viewing/msg6088485/#msg6088485

Thanks for those Python links. I have so far got the thing running under Debian 13 in a VMWARE VM under win10, and the next step will be to move it to a fanless Win10 machine which you can pick up used on Ebay e.g https://www.ebay.co.uk/itm/115710830111 and that can run the VMWARE player v16, or maybe even later. I like VMs because you can move them around.

[paulca]

I like VMs because you can move them around.

Yes, you can.  A word of caution though.  If you are using "consumer grade" SSDs, check the total write budget on the drives.  I trashed a 2Tb M.2 SSD moving proxmox VMs around willy nilly.  It hit its warranty write threshold in under 6 months and it's capacity started to drop.  It was only rated, I believe for 40Tb writes.  Might have  been 400, either way I consumed it all in 6 months of "platform development and experimentation" in a "how the cloud works deep dive".

I now move things sparingly... and I spend twice as much on SSDs with far higher cache and write thresholds.  Samsung EVOs I believe.   If you pay twice again you can get enterprise grade SSDs which very high tolerances and speed.

[peter-h]

Why would a VM trash an SSD faster?

Sure an SSD has a "total write data" limit, somewhere in the terabytes region for a typical 256GB SSD, based on say 10k writes for every 4k block. That is probably why winXP trashes SSDs because it does a write to the same block, at 1Hz. WinXP SSD machines running 24/7 last just 1 year - IME. (It may be ok if in a VM on say a win7 or win10 host, again IME).

[paulca]

Why would a VM trash an SSD faster?

Sure an SSD has a "total write data" limit, somewhere in the terabytes region for a typical 256GB SSD, based on say 10k writes for every 4k block. That is probably why winXP trashes SSDs because it does a write to the same block, at 1Hz. WinXP SSD machines running 24/7 last just 1 year - IME. (It may be ok if in a VM on say a win7 or win10 host, again IME).

Its just the size of the things you are moving around.  Soon it's not just OS VMs it's storage disk images.  Testing things like parallel running and fail over requires a lot of snapshots gets sent back and forth etc.  It sounds far fetched, but it can add up to many, many terrabytes over months.

My Windows DevVM is 100Gb.  The linux one is 80Gb.  Although most server instances are in the order of 2Gb.  I have a couple of large data images for the migratable NFS server for example which serves linux home dirs etc.  When you want to shuffle things around between half a dozen physical hosts you can get a bit carried away.

Actually as you stray further into the virtual platform you will find that storage will become a bottle neck.  While you can of course share real disks from a VM it is often easier to virtuallise the disk itself. Not least as it then becomes a standard backup item automatically.  So it is backedup by incremental snapshot running nightly and a full shutdown and mirror copy weekly.  Of course those features via the VMWare route are expensive.  With QEMU or bare supervisor you have to do all that yourself.  Proxmox is where I settled.  It's annoyingly opinionated and you need to disable a lot of it's crap to make it "consumer useful", but it is stable and provides a decent enough amount of functionality.  It is however a "bare metal install", so Proxmox takes over the whole box it becomes a VM/container Host only.  It's FREE and opensource.

At the extreme end you have "Open Stack" which is a set of tools and components for building your own cloud inftra from scratch, aka AWS, AZure etc style.  Not "out of the box" in any fashion at all though.

[paulca]

What that looks like:


This is just the main "dev server".  pve6 is the real server, a tiny little HP Elitedesk G300 with 32Gb of RAM.

What strike me as odd though.  How can the two mirrored SSDs have different wear out values?  They were installed together using the RAID style failover first one, then the other.