Oberon: what is it good for?

The first version of Oberon was created at the end of 80-ies and the beginning of 90-ies. Niklaus Wirth was back then a well-known and influential figure in Computer Science. His work was open source, well described, and available for download. But time went by and the system did not leave the academic world and his university. Even the German-speaking world seems not to pick it up. So, why? And why should we even pay an attention to this system?

First of all, what is Oberon and what is its goal? It was created as an illustration for the Operating Systems course and exercise in building an operating system that can be fully understood by a good student. The whole system was built originally by one man from scratch. So, it is fantastically compact. It allows for extension, and simultaneously it also allows for the reduction of all the code that is not currently used. It is amazingly reliable. It is lightning fast. But it is not widely used and not even widely known.

It seems to have its influence. People who were creating Windows 3.11 seems to be familiar with the work of Wirth. Creators of the Go language reference it numerous times and it seems to be the major inspiration for the language.

But what can we learn from it now? Where can we use it? And what can we use it for?

What can we tell about the system and team who will utilize these OS and language?

  • A person who will write code on it should understand the system thoroughly. It is as different from other systems (Windows, Linux, Mac Os) as these systems are different from one another. That first step requires an investment of brainwork and habit building and it seems to contribute to the unpopularity of the system.
  • The Oberon system seems to repel porting code from other platforms. so, whatever system will be built on it should be built solely or for (very) most part on the Oberon language. 
  • Whatever is built on the Oberon promises to be extremely fast, small, extremely reliable, and can be supported by a small stable group of knowledgeable men (maybe even by one person).
  • It seems to be extremely hacker-proof from the outside world for a multitude of reasons. It is unknown to uneducated folks, it was built with all the external ports closed, it is well thought through and logical.
  • Seems to be easily portable to new processors by a small team or, even a single person. Already has Arm and x86 compilers. 
So, what are those projects with such characteristics?

Real-time projects: any of them from a simple "smart" lightbulb to sophisticated avionics - including military, robots, and factories. They require all of the above. The only part of the original Oberon that needs to be rebuilt is the scheduler. Anything close to hardware should benefit from the adoption of Oberon. 

Specialized docker sidecars - think about Istio service mesh. Oberon already has all the libraries necessary to build such a system.

Bootstrap for new processors - every time new architecture is introduced new systems are trying to take a place under the sun while people are porting old systems to new platforms. Oberon is ideal to be ported - to something like RISC V.

National programs - several countries are seeking technical independence right now - especially for real-time stuff. Russia, China, Iran, to some degree India, maybe even Japan, and South Korea should be highly interested in a national platform based on Oberon. It can be fully designed, developed, and supported by a small high-quality team.

Maybe it can be used as a base for a smartphone system similar to Android. Android is based on the Linux kernel with Java VM on top. So, that Linux kernel might be replaced by Oberon (with a new scheduler) - despite I am not really sure why to bother.

So, what pet hobby projects can be done to push some of these applications?
  • buy a Zigbee dev kit, and rewrite it on  Oberon.
  • remote client for Oberon (think about RDP for Windows).
  • port Oberon (as-is) on Raspberry Pi
  • To start Oberon on Docker - with the remote client mentioned above 
  • sshd on Oberon, allowing for Oberon shell
  • make a code-generator for Oberon compiler to RiscV, port the whole system to some small RiscV single board computer
  • port ZeroMQ to Oberon
  • port logical library to Oberon.
Will see what I can come up with in the next year. I will also collect links to other people work toward these goals.

Comments

Post a Comment

Popular posts from this blog

Code Reggae

What for?  The blog is intended for using rust programming language   for liberating code ideas that sometimes disturb my sleep. Rust is an "ML in the C skin" language, it is a system-level language allowing for writing an operating system on bare metal (see Redox ), increased reliability (due to ownership model ), rich libraries set, great build system and without the garbage collection. The main idea is to rewrite a well-known piece(s) of software on rust and correct the shortcomings by using the rust unique language properties. Correct what? The main idea is to use rusts ownership model for detection and correction of shared variable usage. Often such usage is scattered throughout the code tree of old and respected C projects and prevents to switch to modern concurrency model. For example, PostgreSQL starts a process (!) per client instead of utilizing threads or co-routines. Data between those processes is shared using OS-level shared memory. That's ugly and...
Read more

Active Oberon is still alive

 After the previous post, I was looking at Oberon-related stuff. Seems that a couple of ideas are still alive - Active Oberon, both operating system and language.  There is an unofficial GitHub mirror of the  Active Oberon working group. Active Oberon team still uses svn, so the unofficial mirror is updated only from time to time. The Github mirror has several builds - I tried the linux64, and everything seems to work just fine.  I even found there an ARM version of the compiler and Active Oberon for an ARM-based board . And this is extremely interesting - because this world does not have a good real-time OS for "embedded" programming. If Oberon can be used for this purpose - it can be a  success. But in order to be successful, someone should port several utilities (well - the simplest version of shell and sshd comes to mind first) and, probably, several C libraries to A2OS (as they call the Active Oberon OS now).  It seems to be really ready for cross-co...
Read more