Free95 claims to be a GPL 3 Windows clone, but it's giving vaporware vibes
- Reference: 1741777272
- News link: https://www.theregister.co.uk/2025/03/12/is_free95_real/
- Source link:
The project description of [1]Versoft's Free95 says that "Free95 is an open source Windows-compatible operating system." Well, not yet it's not. It appears to be barely even a sketch of a demo. The real question is, perhaps, will it ever be what it claims? On this, we are inclined to listen to the [2]advice of noted design authority Carlton D. Ridenhour.
Developer Kap Petrov [3]announced his project on Reddit. There he has received considerable acclaim for his efforts, and at the time of writing, 330 stars on GitHub. The [4]Adafruit blog seems impressed, and it's also been [5]reported on Hackaday . All this for a project that doesn't even have a license yet.
[6]
We confess that we are a little surprised by how well the project is being received, in terms of uncritical news stories, positive comments, and people expressing their hope that they might be able to run Windows apps on a different OS.
[7]
[8]
After all, you can already do that, and have been able to for years now. [9]WINE is pretty good now and can run 16-bit and 32-bit apps on 64-bit OSes, x86 apps on Arm, and more. [10]Including AAA games . And if you want a whole Windows-compatible OS, that exists too, in the form of the [11]now multicore-capable ReactOS .
We find it all the more surprising since there is not a lot to see except two low-quality screenshots and a relatively tiny amount of source code. We can't really judge the quality of the C code, but we can judge this less than compendious [12]remarks file . You can judge the open source credentials of [13]dllme.txt for yourself. There was also a file called malware.c , but it's been removed, which is probably for the best.
[14]
Some other forums are slightly less charitable, but still, not to the degree that you might expect. For instance, from Hacker News…
[15]Asveikau :
Sorry this is going to be more impolite than most HN comments, but this person doesn't know what they're doing.
[…]
This is not what a replacement Windows kernel looks like.
Honestly, that seems very polite to us. The denizens of Slashdot are [16]a little more skeptical . Among our favorite assessments so far are:
[17]Sosume :
It runs only a dos box and notepad, but the developer pinky promises to keep up the work!
[18]Blugalf :
With all due respect to anyone's efforts, but this is as far removed from being a "New Open Source Windows-Compatible Operating System" as a firecracker taped to a wooden stick is from a Falcon 9.
[19]Fresh Wine-flavored version of Mono released
[20]Stuff a Pi-hole in your router because your browser is about to betray you
[21]Essential FOSS tools to make macOS suck less
[22]101 fun things to do with a locked Kindle e-reader
It has taken ReactOS nearly 30 years to get where it is now. The project launched in 1996, under the almost suspiciously similar name of the [23]FreeWin95 project . Trying to reproduce Windows NT is a vast undertaking and we remain concerned that if it ever gets to the point where it's genuinely competitive, even with an elderly version of Windows – such as what we consider the classic version, Windows 2000 – then the full legal wrath of Redmond will descend on it, leaving nothing but a smoking crater.
Just to be clear, we're not saying Free95 is no good, or doomed to failure, or anything like that. We're not questioning its author's abilities. What we are questioning is whether Free95 is in fact an attempt to write an OS at all. We suspect that it started out as a joke, maybe the sort of thing that we gather is now called [24]vibe coding – just freewheeling with an LLM bot coding assistant, improvising, and seeing how far one can get and still have the thing compile successfully, never mind actually do anything useful. But what may have been a bit of fun has gotten rather out of hand and is still rapidly snowballing.
Of course, Petrov could just drop it and move on. He's left comments on Reddit saying that he has a local copy of a version 0.2.1 but he hasn't uploaded it yet – which is not the way that Git works for version control, but is hardly important.
[25]
It has just surprised us how many people seem to be unquestioningly accepting this, without stopping to ask if it's real, or if it's been done before, or, indeed, asking much at all.
We will leave the last word to another Slashdot commenter.
[26]Laughingskeptic :
Out of the 400+ functions listed by Microsoft for WinUser.h he has 2 implemented: FillRect , PtInRect . It is a little premature to be discussing releases and how lightweight this implementation is.
®
Get our [27]Tech Resources
[1] https://github.com/Versoft-Software/Free95
[2] https://www.youtube.com/watch?v=9vQaVIoEjOM
[3] https://www.reddit.com/r/osdev/comments/1j3y9p7/i_built_an_os_to_be_compatible_with_windows/
[4] https://blog.adafruit.com/2025/03/10/free95-a-windows-compatible-operating-system/
[5] https://hackaday.com/2025/03/10/freeing-windows/
[6] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=2&c=2Z9G9tVT_NBH7OIo9fHs50QAAAdg&t=ct%3Dns%26unitnum%3D2%26raptor%3Dcondor%26pos%3Dtop%26test%3D0
[7] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z9G9tVT_NBH7OIo9fHs50QAAAdg&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[8] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z9G9tVT_NBH7OIo9fHs50QAAAdg&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[9] https://www.theregister.com/2025/01/24/wine_turns_10/
[10] https://www.theregister.com/2023/09/27/osseu_steam_os_3/
[11] https://www.theregister.com/2022/03/14/reactos_smp/
[12] https://github.com/Versoft-Software/Free95/blob/main/free95/src/remarks.txt
[13] https://github.com/Versoft-Software/Free95/blob/main/free95/src/tools/dllme.txt
[14] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=4&c=44Z9G9tVT_NBH7OIo9fHs50QAAAdg&t=ct%3Dns%26unitnum%3D4%26raptor%3Dfalcon%26pos%3Dmid%26test%3D0
[15] https://news.ycombinator.com/item?id=43317129
[16] https://tech.slashdot.org/story/25/03/10/0142235/new-open-source-windows-compatible-operating-system-released
[17] https://tech.slashdot.org/comments.pl?sid=23631395&cid=65222557
[18] https://tech.slashdot.org/comments.pl?sid=23631395&cid=65222653
[19] https://www.theregister.com/2025/03/11/new_wineflavored_version_of_mono/
[20] https://www.theregister.com/2025/03/08/pi_hole_6_flyby/
[21] https://www.theregister.com/2025/03/07/fosss_to_tame_macos/
[22] https://www.theregister.com/2025/03/06/kindle_and_modos/
[23] https://reactos.org/wiki/FreeWin95
[24] https://simonwillison.net/2025/Mar/6/vibe-coding/
[25] https://pubads.g.doubleclick.net/gampad/jump?co=1&iu=/6978/reg_software/oses&sz=300x50%7C300x100%7C300x250%7C300x251%7C300x252%7C300x600%7C300x601&tile=3&c=33Z9G9tVT_NBH7OIo9fHs50QAAAdg&t=ct%3Dns%26unitnum%3D3%26raptor%3Deagle%26pos%3Dmid%26test%3D0
[26] https://tech.slashdot.org/comments.pl?sid=23631395&cid=65223819
[27] https://whitepapers.theregister.com/
"It is a little premature to be discussing releases and how lightweight this implementation is."
Nah, I can for sure say that currently it is _very_ lightweight.
"just freewheeling with an LLM bot"
The strcat() code in free95/src/string.c
while (*dest != 0) { *dest = *dest++; }
wouldn't exactly be disconfirmatory.
Re: "just freewheeling with an LLM bot"
Hey, has that LLM been scraping my code?
Wouldn't it be easier...
...to start from scratch, or just fix Linux so it isn't such a pain in the arse.
Re: Wouldn't it be easier...
Or contribute to ReactOS.
Re: Wouldn't it be easier...
> Or contribute to ReactOS.
Exactly so.
How about people use these amazing "AI" coding assistants to demonstrate cleaning up the codebase of some significant old project? All of Symbian is on Github, for instance.
If those LLM bot tools can do half of what their evangelists claim they can, maybe IBM could open source Workplace OS/2 -- Microsoft had zero direct input into that one -- and there could be a FOSS OS/2 clone.
OSfree never got anywhere:
https://www.osfree.org/
Just like FreeVMS although that got _slightly_ further.
https://www.pvv.org/~roart/freevms.html
Re: Wouldn't it be easier...
What's the old saying?
"Everyone wants to create, but nobody wants to maintain. There is no glory in maintenance."
Needless to say, a very large Achilles heel in our modern tech world.
Re: Wouldn't it be easier...
Everyone likes ice cream but no one wants to milk the cow.
Re: Wouldn't it be easier...
Except Dave, and honestly, he seems way too excited about milking the cows, we should probably keep him away from them...
Hey; remember Windows 95 in a web browser?
[archive.org/details/win95_in_dosbox]
> Hey; remember Windows 95 in a web browser?
Yeah, but that was the entire real thing, in an emulator, wasn't it?
As opposed to crazy websites like this:
https://dustinbrett.com/
Which AFAICT is a humongous lump of Javascript pretending to be a desktop OS inside a browser tab. Seems totally pointless to me, but what do I know?
Has 2 working functions, Hey BillG3, he has 2 more than you ever wrote and 1 more than Windows 11 and that's for the phone home telemetry.
The first computer (ENIAC) was invented in 1946. That means it took 49 years to write Windows 95. Even if this is real it's not going to finished till 2074.
AI will finish it while inventing time travel and it was all finished last week. Unfortunately, time travel creates an alternative reality we can't see.
The first was a [1]few years earlier , it was all a bit hush hush though....
[1] https://www.theregister.com/2015/02/20/colossus_tommy_flowers_get_stamped_lorenz_code_breaking/
meh. The first *real* computer was the Manchester Baby:
https://en.wikipedia.org/wiki/Manchester_Baby
meh. I think you're all forgetting about the Antikythera mechanism - which predates those by about 2000 years.
To quote [1]wikipedia "The Antikythera mechanism is generally referred to as the first known analogue computer."
[1] https://en.wikipedia.org/wiki/Antikythera_mechanism#Origin
This should be a Rust project. Perfect vaporware.
But then you'd have to emulate the Win95 memory holes in Rust! That is a challenge.
Who's game to implement a custom malloc surrounded by unsafe, just to hear half the fanbois squeal and the other half saying it's much safer now in Rust.
Just for the record, ReactOS has been worked on for at least a decade and probably longer, and is still a buggy alpha mess. This is going nowhere.
So it's got feature parity with Vista and Windows 8 then.
Even this article gives the project too much deference.
Implementation of PutChar before somebody on Reddit very slowly explained the idea of testing ranges of values:
```
UINT32 index;
if (ch == ' ')
{
index = 26 * 16;
}
else if (ch == '!')
{
index = 27 * 16;
}
else if ... etc for another 42 if statements ...
```
Author's excuse: "Be aware that i wrote this code while i was multitasking and i am terrible at multitasking. Thanks for the idea, and i will surely rewrite that portion of the code."
No matter what his potential, right now this seems to be an enthusiastic child running fast towards the buffers.
But... why?
Why even bother making such a thing?
Novelty is cool, but Win 95 never had solid utility. I was a bridgehead, so speak, for M$.
NT was the real play back then and 95 was just a way to get the average person familiar with the new GUI and general NT workflow.
So, I'll pass just on general principle, thank you very much.
Re: But... why?
I cannot speak for the original author, and think we might be grossly overestimating him if we assume any coherent rationale, but the objective of recreating a Windows 95-compatible version of Win32 would allow a lot more in terms of era-relevant entertainment content.
Windows 95's 'incomplete' runtime fault detection acts as accidental fault tolerance for many titles — in Windows 95 passing a NULL here or there to DirectX gets a free ride whereas under NT you'd get the page fault that one would ideally hope for.
Fantasy
Absolute fantasy.
Why?
Windows 95 was just reskinned Windows for Workgroups, which was mostly a GUI layer on top of DOS with a hodgepodge of 16- and 32-bit components. Fair play to Microsoft, having 32-bit components in that strain of Windows was an innovation, but the world has moved on quite substantially, and the Windows 9x lineage died with Windows Me.
Hopeless
I had a quick look at string.c and found this comment for strcmp.
// Compare two strings. Should return -1 if
// str1 < str2, 0 if they are equal or 1 otherwise.
And the code returns 0, or 1, but not -1.
Remarks file
If that remarks file is anything to go by, they could set up a swear jar and the project would be self-funding in no time.