I smell vapor ware coming. - No one will care about a new OS, if it doesn't run on a wide range of hardware. Tons of examples around. - The file access issue mentioned here or in the presentation is already fixed with flatpak or snaps. - Yeah, some micro kernel with pluggable drivers might be nice or cool or whatever. But it will likely need an evolution out of Linux for that, reusing all the drivers there. - Jesus, this title is so clickbait!
MacOS is popular and it just runs on macs. Not only that, this computer is pretty cheap. It if it ends up decent, I don't see why it wouldnt be viable.
It is so absolutely not fixed by current Linux sandboxing package managers. The tight integration with the OS is crucial. It's important that it is not optional. Also, the current Linux solutions suck. I don't remember if it is installed with flatpak or snap, but my Discord for instance, can't access any files that aren't in my downloads folder for some reason. If this was Serēncord, it would prompt me to give it the file, and it would just work. Because the program doesn't just expect to be able to read anything from anywhere.
@@miguelborges7913 Yeah, and there is OS/2, BeOS, SailfishOS and whatever ton of OS I forgot. Just picking one success story and not looking at the hundreds of failed attempts is not really helpful.
@@hansdietrich1496 While you're not technically wrong, it's also ... not helpful ? - BeOS has a spiritual successor in HaikuOS so in a way it's still alive. Still very niche, but seems to be somewhat useful / daily driveable - a new OS that doesn't run on a wide range of hardware - the water is wet, you know that ? That is true for every OS. What you're telling is that basically noone should start writing an OS because initially it will only run in a VM and/or a single platform with just a few peripherals accessible. If the strong points of the OS are desirable enough people or even companies will pick it up and add hardware support. Whatever drivers are open source can be ported to another OS, especially if it's easy to program in (like this one tries to be). Of course it's not easy in general, few people will do that, but it can totally piggyback from Linux' drivers, so I don't see this as an actual hard problem I do agree that the titles and thumbnails are massively clickbaity. Funny at times but always exaggerated.
@@miguelborges7913 This computer is actually not cheap. Especially for how weak the hardware is (RISC-V CPU with DDR3 or DDR2). But it is how Sam can pay rent and electricity and stuff, so I'm not against it. I do am excited to see how well it develops. If it's able to make it run as well as, say, N100 with Linux on this inferior hardware. That would be a point on clean, lean and powerful kernel / OS.
OK, go ahead and create your own perfect OS vision, but keep in mind that other OS developers are not stupid. They had reasons for every decision that led to the current state of the Linux operating system.
@@no_name4796 Well all OSes are UNIX clones for compatibility. Imagine Linux introducing this security model, every program wanting to read a file would break, maybe you could design your way around this by running older programs in do-everything-lol-compatibility-mode, but how do you convince everyone to use the new thing, maybe you want non copy-on-write memory and not duplicate the entire address space for every newly forked process, so you axe fork, that breaks a lot of programs (unless they use clone, rfork or something). Maybe you think ioctl and signals are bad, so you want to remove them for Linux, you can't because programs rely on that. My point is, other OS developers cannot innovate, we are stuck with UNIX, until something better comes along. But this Serenum thing is just an overpriced SBC with a bare-bones OS that is not really useful to anyone, other research OSes are also relatively niche. Closed hardware and proprietary graphics driver make OS innovation and adoption slow (writing and marketing an OS was a lot easier when everything was standardised to the IBM compatible PC, thats why so many OSes come from that time period) see the thirty million line problem by Casey muratori why there are no relevant new OSes. Though there are some promising new innovations, e. g. fuchsia from Google, or Oxide Hubris System those are large enough companies and they funding to be able to create a successful new OS.
They are not stupid, but also didn't saw a "let's start from scratch" option as being viable. Because they are / were alone and without a massive time and budget, so all of their talent went into fixing / improving existing OS, which inherently has limitations.
If i was to make an OS, I would allow the exe file access all files in the same folder or below, and then only allow program to run from a programs folder. If it needed to access other files then it would need extra permissions. But that should be rare. Programs should be "contained" in their own environment.
There's sandboxing ,in the bsd's which is unix there's jails and in qube os which is by the way a linux distro you can spawn ' a qube ' for any program you want and incase you get hacked you can spawn another cube. So?
Firstly, Linus didn't have an OS to sell and secondly, had his kernel not been free, GNU would never have picked it up and you would have heard of neither Torvalds nor Linux today. Poor comment.
It's such a shame that we can't have nice things (all the ideas you're implementing in your OS) bc business is more concerned with profits than actually developing cutting-edge technology. And they do this by running outdated, unsecure, legacy code and ideas from the 70s, 80s and 90s.
does this has to be an own OS? can't that functionality simply be embedded in an executable. E.g. if you want to run Craftmine securery, then you use a wrapper, like `runsec /usr/bin/craftmine`
I see two key problems with this: - It is important that it is not optional. - Craftmine would be written in a way expecting to be able to access the entire file system in this case, and would not call out to the WM file manager.
this is a very interesting idea.
this is absolutely not "Linus Torvalds' Big Mistake" in any way.
the clickbait on this platform really is the worst
I smell vapor ware coming.
- No one will care about a new OS, if it doesn't run on a wide range of hardware. Tons of examples around.
- The file access issue mentioned here or in the presentation is already fixed with flatpak or snaps.
- Yeah, some micro kernel with pluggable drivers might be nice or cool or whatever. But it will likely need an evolution out of Linux for that, reusing all the drivers there.
- Jesus, this title is so clickbait!
MacOS is popular and it just runs on macs. Not only that, this computer is pretty cheap. It if it ends up decent, I don't see why it wouldnt be viable.
It is so absolutely not fixed by current Linux sandboxing package managers. The tight integration with the OS is crucial. It's important that it is not optional. Also, the current Linux solutions suck. I don't remember if it is installed with flatpak or snap, but my Discord for instance, can't access any files that aren't in my downloads folder for some reason. If this was Serēncord, it would prompt me to give it the file, and it would just work. Because the program doesn't just expect to be able to read anything from anywhere.
@@miguelborges7913 Yeah, and there is OS/2, BeOS, SailfishOS and whatever ton of OS I forgot. Just picking one success story and not looking at the hundreds of failed attempts is not really helpful.
@@hansdietrich1496 While you're not technically wrong, it's also ... not helpful ?
- BeOS has a spiritual successor in HaikuOS so in a way it's still alive. Still very niche, but seems to be somewhat useful / daily driveable
- a new OS that doesn't run on a wide range of hardware - the water is wet, you know that ? That is true for every OS. What you're telling is that basically noone should start writing an OS because initially it will only run in a VM and/or a single platform with just a few peripherals accessible. If the strong points of the OS are desirable enough people or even companies will pick it up and add hardware support. Whatever drivers are open source can be ported to another OS, especially if it's easy to program in (like this one tries to be). Of course it's not easy in general, few people will do that, but it can totally piggyback from Linux' drivers, so I don't see this as an actual hard problem
I do agree that the titles and thumbnails are massively clickbaity. Funny at times but always exaggerated.
@@miguelborges7913 This computer is actually not cheap. Especially for how weak the hardware is (RISC-V CPU with DDR3 or DDR2). But it is how Sam can pay rent and electricity and stuff, so I'm not against it.
I do am excited to see how well it develops. If it's able to make it run as well as, say, N100 with Linux on this inferior hardware. That would be a point on clean, lean and powerful kernel / OS.
OK, go ahead and create your own perfect OS vision, but keep in mind that other OS developers are not stupid. They had reasons for every decision that led to the current state of the Linux operating system.
Yup.
There are examples of people building a big project, getting agry about how shitty it is, then they rewrite it just to get the exact same thing.
@@no_name4796 Well all OSes are UNIX clones for compatibility. Imagine Linux introducing this security model, every program wanting to read a file would break, maybe you could design your way around this by running older programs in do-everything-lol-compatibility-mode, but how do you convince everyone to use the new thing, maybe you want non copy-on-write memory and not duplicate the entire address space for every newly forked process, so you axe fork, that breaks a lot of programs (unless they use clone, rfork or something). Maybe you think ioctl and signals are bad, so you want to remove them for Linux, you can't because programs rely on that. My point is, other OS developers cannot innovate, we are stuck with UNIX, until something better comes along. But this Serenum thing is just an overpriced SBC with a bare-bones OS that is not really useful to anyone, other research OSes are also relatively niche. Closed hardware and proprietary graphics driver make OS innovation and adoption slow (writing and marketing an OS was a lot easier when everything was standardised to the IBM compatible PC, thats why so many OSes come from that time period) see the thirty million line problem by Casey muratori why there are no relevant new OSes. Though there are some promising new innovations, e. g. fuchsia from Google, or Oxide Hubris System those are large enough companies and they funding to be able to create a successful new OS.
They are not stupid, but also didn't saw a "let's start from scratch" option as being viable. Because they are / were alone and without a massive time and budget, so all of their talent went into fixing / improving existing OS, which inherently has limitations.
Never question the old gods. Just use their design and never change anything. lol.
@@instcake You can question anything, I do it all the time. But assuming you are smarter is not a smart move.
Simply have no bugs
wake me up in about two years when this os has died
"Native is the only option" 🔥
12:12 You have Firefox and Qt on your own OS?
No they were temporary entries for testing the run menu.
you can use firejail to wrap binaries that you dont trust.
If i was to make an OS, I would allow the exe file access all files in the same folder or below, and then only allow program to run from a programs folder. If it needed to access other files then it would need extra permissions. But that should be rare. Programs should be "contained" in their own environment.
so make one...
Is Flatpak sufficient in your opinion?
@@khai96x yes if it wasnt trash
This is just mount point namespace + the file picker portal
There's sandboxing ,in the bsd's which is unix there's jails and in qube os which is by the way a linux distro you can spawn ' a qube ' for any program you want and incase you get hacked you can spawn another cube. So?
qubes isn't a linux distro
@@angelcaru then what is it? Go to their site and look what qube os is
Linus Torvalds' big mistake was not selling his OS exclusively on $350 shitboxes flipped from Aliexpress. What a joke.
Firstly, Linus didn't have an OS to sell and secondly, had his kernel not been free, GNU would never have picked it up and you would have heard of neither Torvalds nor Linux today. Poor comment.
Do you plan to have shared libraries? Because that would mess up the reproducibility
No. Shared libraries are explicitly not allowed.
I see no tech specs? wtf....
Tell me you never used Mandatory access control without telling me you never used Mandatory access control...
I'm completely lost in this discussion
RBAC makes more sense to be fair.
Look at the insanity that is Active Directory or Windows, or SELinux.
IRL everyone manages access based on roles.
I don't think you understand, it is at the core of the OS, not a addon thing.
Care to elaborate ?
It's such a shame that we can't have nice things (all the ideas you're implementing in your OS) bc business is more concerned with profits than actually developing cutting-edge technology. And they do this by running outdated, unsecure, legacy code and ideas from the 70s, 80s and 90s.
Who said we can't.
eh not really
does this has to be an own OS? can't that functionality simply be embedded in an executable.
E.g. if you want to run Craftmine securery, then you use a wrapper, like `runsec /usr/bin/craftmine`
I see two key problems with this:
- It is important that it is not optional.
- Craftmine would be written in a way expecting to be able to access the entire file system in this case, and would not call out to the WM file manager.