Bob Washburne is the first to use the new "rant" section on ANN v2, intended as a kind of soap box for normal users to voice their opinion and perhaps get some people thinking. Anyway, please tell him what you think of his writing, and please tell me what you think of this concept. :)
There has been a lot of confusion in the past several months on just what the Amiga community wants in a new computer and, in fact, just what an amiga is. I hope the following thoughts will help to clearify the issue and perhaps even inspire some action.
What is an Amiga? An Amiga is one of those computers I have down in my basement. They have numbers on them such as 2000, 2500 and 1200. They consist of hardware, software and an OS. This is an Amiga. Period.
An Amiga is *NOT* a concept, a spirit, a community or a Way of Life (tm). All of us bought our Amigas because they ran really neat software really well. We continue to use our Amigas because they still do.
What do current Amiga users - the Community - want to see in the next generation box? We simply want a computer which will 1) run our current software as well as or better than our current Amigas. 2) allow us to use modern, i.e. fast and cheap, peripherals such as with IDE and PCI. 3) Have features which will bring us more neat software which can't be found on other computers.
What about the new computer being worked on by Escom/Amiga, ermmm, Gateway/Amiga, ermmm, Amino/Amiga? The facts of life are that any startup company must build a computer with sufficient appeal to make the company viable. By the time you do your marketing study you find that any computer design which is sufficiently popular will have no commonality with the current Amiga line. It may be nice, even neat, but it will be an Amiga in name only.
What about QNX? Nice kernal, but nothing Amiga here. They have been working on their desktop environment for quite a while, long before Collas ever approached them. They have offered to take their existing project and place the Amiga name on it, but that is the only thing Amiga about it - no software compatability and none planned. And QNX has possibly less support for hardware than the current Amigas do. Just try and find a web site or news group where people are busily writing drivers for QNX.
What about Linux? When you make a list of all the features missing from AmigaOS; shared memory, memory protection, SMP, etc., you pretty much come up with the specification for Linux kernel 4.x. And Linux would not be a bad choice to replace exec. BUT, that would take a lot of time and work. It has taken hundreds of programmers almost ten years to bring Linux to its present state. It will be at least another year to incorporate the rest of the features mentioned. Then you would need to rip the kernel appart to remove stuff not of interest to Amiga and then completely rewrite Workbench to run on this new kernel (And removing all refferences to the propietary chip set. So long, Agnes). Bottom line: it would take a large programming team at least two years to replace exec with Linux and still be software compatable with the current Amiga. A company as small as Amiga, Inc. cannot bring the resources needed for this project. And even if they did, it would not pay off for them.
What if AmigaOS is made open-source? Nice idea, but many problems;
What about UAE? Great idea, if only it worked. Remember, the Amiga has several independantly programmable processors; CPU, copper, Agnes. etc. To get a single processor machine, such as Intel, to emulate this requires a LOT of horepower. My 233MHz AMD cannot fully emulate a 7MHz A500. I suspect that it will take at least a 1000MHz machine to emulate the A1200 fully. And then what do you have? A chance to preserve the past, but no path for the future.
- Some of Workbench is covered by patents which are still owned by Gateway. So you can't just GPL Workbench.
- You still need the massive rewrite to get around the OCS, ECS, AGA issue.
- There are only so many geeks out there and for the most part we are all working on Linux and FreeBSD. How are we going to attract them to Amiga in sufficient numbers to make a difference?
- Who is going to be in charge? There must be a kernel czar who decides what goes in and what does not. Otherwise you end up with many different, incompatable kernels. Linus Torvalds and Allen Cox do this for Linux. Who will do it for Workbench? Amiga, Inc.? (see above) Phoenix? H&P? COSA? Petro?
Is there any hope at all for a *REAL* Amiga NG? Surprisingly, yes. Transmeta. The Caruso processor chip is supposed to be fast (400-700MHz), cheap ($70-$130 US) and able to directly run the binary code from any other processor without emulation. So all it would take would be for some enterprising company to design a motherboard with multiple Caruso chips. One then becomes the 680x0, one becomes Agnes, copper, Paula, etc. You would then have a computer which is fully binary compatable with the current Amiga line, but runs a 400/700MHz CPU with a 400/700MHz AGA chipset.
This would be a great leap forward for Amiga technology. And if the board was designed well, you could also sell it as an SMP Linux system. This would provide the market volume necessary to pay back the development costs and then some.
And it would be a flexible platform from which the Amiga could move forward. Open sourceing Workbench makes sence at this point as now, with the performance pressure off, we would have time to rewrite the proprietary stuff out and the new features in.
So, any comments?