General discussion


Time for a new desktop OS

By The Admiral ·
Over the past 10 years, I have seen a certain disdain for any desktop operating system other than Microsoft. Windows 95 kicked it off with making the user interface more ?intuitive? than many of the other operating systems. At the time, many of the other operating systems that were being developed started to look more like Windows with the GUI. The problem with those OS?s has always been trying to get the hardware vendors to sit up and take notice.

OS/2, one of IBM?s more stable Desktop OS products was initially in the running with the 32-bit architecture, being the first to market with the OS to desktops that were still running Windows 3.1. Jointly built by IBM and Microsoft, it had all of the great benefits of becoming the next Windows NT, until the two companies split. Once that tear happened, the Operating System war started, and the fallout from that war is still being felt.

Now that we see that IBM is pulling out of OS/2, and Microsoft is riddled with problems before the Operating Systems are released, it is time to start thinking about the next evolution of the operating system. The BE/Os was a great step forward, but unfortunately, did not take off with any measurable success in the retail sector. It is time for a division between desktop and server operating systems in that what riddles the desktop OS does not riddle the server OS, as we have seen in previous times with other OS?s.

Computer users around the world are clamoring for a set and forget OS where they can install it in less than half an hour, and when it is installed, it is resistant to viruses, spy ware, and other nasty bugs. I don?t blame them. From my perspective, I am sick and tired of having to keep my definitions updated and the patches streaming because of the seemingly large lack of forward thinking in products. It may not even be the lack of forward thinking, but the lack of intelligent management of problems that lead to the case of a new OS.

I suggest a prom based kernel OS where when you turn the power on, it is there, and there is no problems with the OS. The OS is what drives everything. If you change out the modem, then you lost the ability to use the modem because the OS on a PROM don?t recognize the card, it has to be done through software. Applications can be loaded, but since it will be loaded in its own memory space rather than be integrated into the OS, if the Word Processor Crashes, it does not take down the desktop or other processes working in the background. If you want to do multi-media, since most if not all of the functions is done by software already, why the need to have it integrated into the OS?

I am not being a purist here at all, I just believe that the more junk that we put into an operating system in order to do the stuff we want on demand, the more patches we need in order to keep it running. Who ever heard of needing to patch sound drivers because of an exploit? Or even a better one, a printer driver that allows a hacker to gain access to a system, or a manner of rendering the graphics from a webpage allows someone to take over your system? It has gone from mildly amusing to lunacy.

I suggest this, because companies are spending too much time and effort on items that should be the vendor?s responsibility. Microsoft should be drawn and quartered if their system (which by the way was patched previously ? remember unPnP?) is still vulnerable. Then an aggregate release of windows with all of the patches goes from Windows 2000 to Windows XP? Please, how about canning the speeches on innovation and start talking about the value that your operating system brings me if it is unbreakable.

For years, the Commodore 64 and Commodore 128 was something I used because of how easy it was to get rid of corruption in the Kernel. Power On/Power Off was the solution. Within 10 seconds, I was sitting there ready for the next program to be loaded and ran. Slow, sure, but those easily learned lessons about computing has become lost in a sea of bureaucracy, and now all users from Mom & Dad to large corporations are feeling the pinch. Just how long are we going to go before we start telling companies like the Operating System makers for every vulnerability in the OS that is found we demand a refund of a proportionate amount of money?

What the heck, if the Anti-Tobacco Lobby can blame companies for people?s bad habits, then why can?t we blame manufacturers of OS?s and applications for bad or vulnerable code?

This conversation is currently closed to new comments.

Thread display: Collapse - | Expand +

All Comments

Collapse -

Picture this

by Tony Hopkinson In reply to Time for a new desktop OS

A bunch of IT CEOs stood before a house committee, serially swearing
'We never knew client side scripting was a security risk'.


Collapse -

You are (in essence) suggesting an exokernel

by jmgarvin In reply to Time for a new desktop OS

While in theory this is workable, the overhead can be a killer. If you can solve these issue, you can make a ton of money:
1) How do we gracefully (and quickly) load and unload "modules" so that there is no peformance hit on the system. Context switching probably won't be workable, so something more "interesting" is needed.

2) How do we keep track of various hardware and software (if the hardware or software changes how do we keep that in memory?) Part of the problem is that the PROM isn't writable (at least I'm assuming).

3) Paging/segmation can be problematic with lots of thrashing as disk access with be more often without proper "module" management

(I have more, but I'm out of time)

Collapse -

by The Admiral In reply to You are (in essence) sugg ...

1. Never was a big problem with 16 Bit OS's. Loading and unloading modules is not a necessity if your Kernel has the hardware and appropriate modules for the OS already pre-configured.

2. Hardware for base systems is not a problem. The software for additional items can be placed in a drivers folder where install programs have access to then is locked down except for system processes.

3. Paging and segmentation is always going to be a problem even with proper module management.

Just remember, the OS is on the Prom, applications are not supposed to run in it. The Virtual stuff is to happen outside of the realm of the prom into seperate processes. Memory space is divvied up by need, and rather than having virtual space, you have space the apps can run in. The Prom is responsible for the stable running of the environment. All of the other virtual items are done outside of the main OS code.

Collapse -

DMA for the PROM

by jmgarvin In reply to

Wouldn't the PROM need at least one DMA specifically reserved for it?

Collapse -

DMA not required - PROM is just where the code lives

by TheMonster In reply to DMA for the PROM

OS code loaded off PROM (probably EEPROM, with a physical switch that must be manually engaged to enable writes) doesn't require a dedicated DMA any more than code loaded from a hard drive or over the network.

As EEPROMs have gotten larger, it has become posible to include LILO/GRUB, a Linux kernel, init scripts, shell, libraries (including X) along with a replacement for the old BIOS. I'd want to have a basic ROM that never gets overwritten, which can in an emergency boot a floppy, CD, HD, or USB stick with repair tools on it, and preferrably several separate banks of EEPROM for redundancy.

I'd use cramfs for most of this job.
A system that can load these things from EEPROM while the hard drives are still spinning up could really be an 'appliance' that can be turned on and off at will, and be especially resistant to viruses because of the write protection.

Collapse -

exokernels can load new modules, its better

by a.mcpherson In reply to

The main difference with exokernels and current microkernels is
that exokernels load modules depending on what the application
needs to run.
The exokernel is even smaller than a microkernel and it is
dynamic, your circumstances are not going to always be the
same, except on embedded systems like traffic lights.
However, if you want a fixed and stable OS, do it on the PROM.
But, I am making my OS able to run on a wide variety of
hardware as a portable CD / DVD / USB key system you can use
anywhere you are. By not fixing the OS to the computer, you
gain continuity of user experience regardless of computer.

Collapse -

trying this on KAOS (bsd)

by a.mcpherson In reply to You are (in essence) sugg ...

I'm doing an exokernel rewrite of the netbsd kernel. This, along
with a new Gui that takes advantage of the new features is called
KAOS. I would have this running on pretty much anything,
including the toaster (yes, there is a toaster running netbsd), the
idea is not to do the OS on PROM unless you won't update the
There are EEPROMs, but it's still a bad idea to implement an OS by
ROM as you lose that adaptabilty you need for hardware updates.

Collapse -

Open source?

by lordshipmayhem In reply to Time for a new desktop OS

I notice you're not mentioning Linux or BSD?

Collapse -

I think you are on to something here...

by rmjrenneboog In reply to Time for a new desktop OS

Whoa, people! I really think you are on to something with this train of ideas. Instead of having a huge, massive OS living as a transient species on your HD, why not have a basic OS (that just tells the hardware what it can and can't do) living in a removable module? Software could then live in individual modules that could be added or removed as needed, according to the tasks at hand. Want to run Linux instead of Windows? Pop out the Windows OS module and plug in the Linux OS module, or run them side by side. Word processing module goes here, spreadsheet module goes there, database module, office suite module, graphics module, etc. etc. etc. Do you see where this is heading? It's been right in front of our faces since the days of the old VIC-20 ind Sinclair Z80s, but we didn't see it or where it could go, and we let Bill Gates and the others lead us down the path of monopolistic profit-driven monolithic design that has brought us to the current mess in which we find ourselves. I say it is indeed time that the original design path was resurrected and brought into the 21st century!

Collapse -

What you want is an exokernel?

by a.mcpherson In reply to I think you are on to som ...

I am making an exokernel rewrite of the Netbsd microkernel.
If you have applications load OS modules before first run, then you
get this very stable OS where a app crash only takes down that app.

Related Discussions

Related Forums