From the start, Linux is sysV and Slackware has bsd commands. But, the reusable codes were rooted in the way bsd does their files(reusable one after another). So, Linux open source codes is to open small file to do each function. Ideally, it works when you have a tiny operating system. If you don't have 500 distros all using the same files but different bindings(bloatedness). Or Linus and Andrew did not release a new kernel version each week(some functions are not universally used)?

All the apps then have to be kernel dependent.

What is not understood is how the open source codes are changed to binaries by compiler(GCC. T2, etc.) versions. So, most of us made a lot of mistakes by using a compiler that had too many files with the wrong bindings, or whichever got compiled into machine language of the wrong platform.

Thus, we have to have computer manufacturers do their own operating system(instant-on preferred, Asus netbook runtime code is just the beginning) to take advantage of every instruction in the set of their cpu. Then we can go straight to the cloud computing on some one's server which is maintained flawlessly.

Thank god. We are also planning using branch prediction(Intel Pentium I) for 8 bit cpu to fake 32 bit or 64 bit functions. We will be able to do multicored 8 bit cpu, more cores then 32 bit cores on the same chip real estate size.

One core will run one OS and time sliced(realtime OS) on running threads. Each core will be suppied with tasks(multitasking operation) by a controller ot hypervisor(a cpu with cache and data inputs such as keyboard and mouse).

Java was open source, then Microsoft took over with their own script compiler. SunMicro sued to get the rights back, now there is only one version of script and its compiler.

Script is english slang, that applies reusable software in the compiler. So, the compiler is on a higher level of compiling files in sequence to make up a huge software.

Linus Torvalds had no control of kernel developers; he does not have a firm working policy to guide open community. So, script compilers such as GCC changes by the wind. Compiler is no longer dictated even by the reference architecture of any computers(platforms of x86 or mainframe computers).

So, now we have to go back to C++ compiler to get function key files of the netbiook by their manufacturers. From there(Javascript), we can do script compilers of browser with its toolbars and taskbars and addressbars?

That browser compiler will then be able to create a new open source Lie-nux operating system? Google wishes to do it with Chrome, starting with C++ compilers to make small files of function keys that conforms to javascript. Then chromium can become a browser of a basic computer operating system for SOA and SAAS internet cloud computing.

It could still be the beginning of a Lie-nux operating system?

As long as we can remember, host being the cpu runs peripherals. This is why system analysis shows computers to be slow. IBMPC architecture is very slow with memory all chopped up into chunks and have to use FAT tables to access them.

When netbooks go with mainframe computer design, the architecture is much simpler. The host just do data transfer. The USB devices can be self contained and respond to data transfer. No more host assisted peripheral device. Lets look at a USB terabyte hdd design to show what can be done.

USB terabyte hdd can have two cpus, each with flash card. Added to several hdds for terabyte capacity, you can have Raid5. Split the hdds, then you can have Raid5 plus Raid1 to make Raid6, which has redundancy safe and secure data storage. You can have banks and banks of USB terabyte hdds connected to your telco or cable MSO servers(PBX) for edgeQam channels.

This of course leads to USB on-demand servers with millions of flash cards or small one inch hdd to download for users from terabyte servers. On-demand servers will relieve denial of service on the net. They are inserted between terabyte hdd and PBX servers with edgeQam channels.

So, the new class of data center will have all new equipment. No more IBMPC type of servers, and no more cisco type of hubs. USB will go from hypertransport pipelines to USB bus with edgeQam amplifiers and to telco servers to the internet cloud.

Since most of the new data centers are designed wrong. It will take one more year to have new data centers installed with new class of equipment emphasizing multi-channel edgeQam simultaneous data transfer of 1024 channels supporting millions of threads.

Ethernet started with PCI bus which ran with 33 mhz in four serial ports. Gigabit ethernet cards lack popularity and not versatile enough to add peripherals in ethernet hubs.

So, we now have USB 3.0(including edgeQam) in the works. But netbooks are designed with mainframe computer architecture in mind. And the USB bus is not only hypertransport pipeline, but also can have multichannel amplifiers to lengthen the HFC(high frequency copper network of wires).

Obviously, netbooks started a revolution of both hardware design and software simplification. People will start to work with netbooks and USB peripherals, such as monitor, printer, terabyte hdd,, etc.

the era of IBMPC architecture will be gone in a few years. Because of the computer simplicity, the cpu, bios and runtime codes are so simple that USB peripheral hotplug can be extremely convenient. Currently, netbooks turn on in 20 seconds, and most apps run withing 4 seconds. There is still room to speed up responses by removing drams and directly accessing the flash card(partitioning the operating system of runtime codes and its swap file from supporting apps and library).

Its time to invent. Its time to change.

Software started on 8 bit ascii codes. So that registers of 8 bits can run ascii codes machine language. Then software had to run 16 bits of memory addresses. Eventually, 32 nits had to be used and banks of memory made up the huge memory control. 64 bit long instructions are often not available to use because of data compression and serial data packet switching methods. Most of the ascii codes handles the modern computers quite well. To run 32 bit software on 32 bit cpu, Intel came up with branch prediction which runs ascii codes better upto 23 stages now in 64 bit cpus.

So, design of computers with massive integration, is divided into many 7 bit cpus handling on either side of a USB bus. Each function requires a cpu to run the device.

So, 8 bit cpu is designed with many pages of memory(protected). The registers may run parallel data at 400 mhz, but the USB will be running at 4 ghz fast. No wait state if possible. Netbooks will move drams either onto the cpu or into the driver of any flash cards.

Ah so. This is a revolution in progress, learn as we go. But the IBMPc architecture will be gone, obsoleted by speed requirement. 64 bit software will never be used because of bitstream serial data which in the digital world is only 2 bits. In the analog world bit streams are video data of maybe upto hundred bits per bit transferred by the conversion of digital bit voltage reading. We will be building analog computers(pure DSPs) soon(beyond smart phones).

It will never be the same again. But then progress is only how we learn to do things; by machines, not by the fragile memory of people's brain.

Footnote:
Ascii code table was invented by Teletype to print out the data transmitted thru teletype machines. It has machine functions(done by asic in cpu). Hardware asic can interpret function keys as embedded operating system. It has alpha numerical data and some graphics characters too for data transfer. It is the cave man's video software instructions but had no sound or speech.

Instant-on is using bios and VGA(791 vesa is much faster than Xorg) to get splash screen(GUI) and use cheat codes to install Linux runtime codes from the flash card. Xinit and Xorg can be used(when applicable) if you get beyond gui panels.

You can cheat by using apps to load xinit and back to VGA by trailer codes(back to gui).

You can cheat to use suspend in bios to make computer green(less power and cpu cooling); activate(resume) with mouse, keyboard and data input from USB, ethernet or wireless.

You can cheat by using DMA to go direct from L2 cache to flash card(hypertransport or hyperthreading pipeline to USB bus) for library dependency(template) files in flash card, without using drams(DDR bus).

You can cheat by using all USB devices, never AGP, PCI, pcmcia or scsi buses.

A few cheat codes will help you speed up Linux?

Footnote:
When you boot, bios gives you a bootloader to do batch processing of your iso file. This is where you have a choice of cheat codes. Which installs Linux the way you like it. Which is going to apps whether on desktop or browser. In each case, then you install peripherals or not, depending on the apps. Cheat codes can run Linux installer, or an apps installer with selective computer functions in library template files(prefetched by L2 cache)?

This means, flash card has permanent(runtime) execution cache of 16 mB(Austrumi has 9 mB); 1/4 gB for template(essential function key execution) files and other library files if desired. /usr has apps for 64 mB(Austrumi has 24 mB), and swap file of 128 mB or more for history of webpages. This a proforma instant-on architecture.

Apparently, Cell phones are now designed like netbooks. Netbooks are struggling to become instant-on. The hardware is merging. Atom chipset has DSP for video. 3G modem is USB.

Android has ARM(DSP) platform. The operating system is based on WindRiver RTOS(packet switching). Then browser(Chromium in the future) took over as desktop. Other cellphone features such as camera and GPS system follows. Making cellphones all in one mainframe computer.

WindRiver can easily do all the netbooks with RTOS and browser. Where will Linux be in the name of instant-on or embedded operating system? Bloated desktop apps that can be done by cloud computing(SOA and Saas built in).

Eventually, Linux will be replaced, even in new data centers. The servers are mainframe design. Buses are USB patched on relay racks with amplifiers. Each server(data storage, on-demand distribution and PBX to users) are all single purpose software. No more open source backward compatible protocols.

The handwriting is on the wall? Instant-on is the design standard; operating system has to comply.