Click for: AFPL GS850 homepage

web page last updated 23rd April 2006,

I've decided to restart this webpage afresh.

project updates: 2006.apr.23, 2006.apr.06, 2006.apr.03, 2006.mar.15, 2006.jan.27, 2005.dec.19, general update

This project is actually several inter-related subprojects.

For many years I've wanted to create my own filesystem (FS). I've been using AmigaDOS and FFS for a long time. AmigaDOS was revolutionary at its time and still is in many ways. eg OFS created a lot of functionality for floppy disks.

The problem was that AmigaDOS had a lot of brilliant ideas but didnt quite go the full distance. As with everything in AmigaOS they repeatedly broke with tradition and created what IMO is the only OS that is nice to use. The user experience is generally well designed, eg "assigns" are the best way to do things. However they put various pointless restrictions into AmigaDOS and FFS. Some of the restrictions were because they used BCPL which eg limits strings to approx 256 bytes by storing the string length in a byte.

I want to go the full distance myself, and create a FS the way I think it should be done with no shortcuts or compromises.

I know that there are various high quality FS's out there, and yes I know you use SFS and have no plan of switching!

:please dont email me this opinion as I've heard it more times than I can count! My apologies, yes you use PFS and not SFS, sorry for mislabelling you. Or you think Linux is wonderful, yes I've heard it all before!

When the project is more mature I will compare it with SFS and any other popular systems, however my project is completely original and I'm not copying any other system. Its too early right now for in depth comparisons.

AFAICS SFS has taken AmigaDOS as a starting point and then moved forwards as far as possible from there, working around the design restrictions.

My project has quite a different intention and aims to replace AmigaDOS rather than continue it. So I completely ignore AmigaDOS in my work and I also ignore all other systems.

I intend to create a redirection interface so that my system can be run above AmigaDOS and vice versa but that will be later, so at that point I will pay attention to AmigaDOS but not now.

I begun the project in July 2004, so its now being going for more than a year.

One of the subprojects was to create my own DVD + CD filesystems, I've already created an optical drive interface for dealing specifically with DVD+RW and CD-RW. I created from scratch a simple + effective partitioning scheme for these media.

Annoyingly I have found DVD+RW drives to be unreliable, the drives will suddenly freeze up. infuriating. So I may restrict attention to CD-RW.

To avoid too many delays I decided to temporarily shelve the optical work and press ahead with the IDE filesystem work. At a later point I will restart the optical work but I dont enjoy coding unreliable hardware.

The nice thing about IDE is that the behaviour is unbelievably reliable.

I achieved a first draft IDE system which creates a FS in any HDToolBox partition. This has been tried out by exactly 3 people: Michael Merkel on OS4, Joe Fenton on a 68k system and also by me.

I got external people to try it for exactly one reason which was to prove the system is for real and not some vapour fantasy. Its much too early though to prove existence to a wider audience.

The first draft system wasnt as fast as I wanted, and after a lot of study I've drawn up elaborate plans for the immediate next phases.

The system I sent to Michael Merkel and Joe Fenton will deal with partitions and files up to 2048G. The design is fully 64 bit, and by changing build parameters I can create files + partitions up to 64 bits.

I'm doing a major overhaul of the work so far, working towards an elaborate caching system. I've been working towards the caching system for some months now. Progress has been slow, partly because of distractions and partly because of motivation problems: some of the required coding is intensely complicated and very technical.

Right at this moment I'm just completing the most difficult part of the new coding. So the project should start accelerating again.

Working on this has made me realise why AmigaDOS didnt go the full distance, I think they must have been working to a deadline and probably needed 4 x as much time as they had to do the work properly.

I havent made any decisions yet which platforms will be supported. All the work currently is on 68k which I run on WinUAE.

The project itself is fully portable and dual endian, to port it to a specific platform wont take tooo long (maybe 3 weeks) but requires knowledge of how the specific platform interfaces with drive hardware. So at the moment it has only been ported to 68k.

For me this is the most difficult project I have ever attempted. It is also the biggest learning experience I have ever undertaken. Its like a university exam that lasts for a year!

Even now I have absolutely no idea how much more time it will take.

The project source is only stored in 1030 bit PGP encrypted format, at the start of a session I decrypt it and at the end I re-encrypt it. That's a security measure to prevent theft by any means.

Because I created the work entirely from scratch I have undivided ownership, and I will never sell the rights. I may consider selling rights to binaries of specific builds but not to the source.

With programming the moment you sell rights to source you will be instantly factored out of the equation. Apparently the author of Internet Explorer sold the rights for a 10% royalty to MS. MS then gave IE for free with Windows and the author thereby got 10% of zero.

Remember: the hand that writes a con-tract is the hand that takes the cash!

conclusion: if you want to contract out your own software make sure you draft your own contract. Dont do deals with companies that give ultimatums to sign a contract as that is a sign of contempt.

Project updates

=========== 23rd April 2006 =========

Further project updates will for now be on my OS project webpage. The FS project is the most difficult part of the OS project.

=========== 6th April 2006 =========

I've got the project to build without build errors. Took quite some work to get it to build. :this completes the action started on 19th December 2005!

When I say a project is sleeping I really mean that. Its no longer sleeping now and is being actively worked on.

Not tried running it, instead I'm studying and improving the code as it currently is.

=========== 3rd April 2006 =========

I've finally restarted the FS project. I will be working on it in parallel to several other projects so progress will be slower. Quite some distance to go yet.

=========== 15th March 2006 =========

I havent done much further with the FS this year. However I've done a lot of x86 asm work preparing the way to using the FS directly on x86 h/w. As soon as I've made sufficient progress on this lower level work I'll resume the FS project.

So the FS project is sleeping currently.

=========== 27th January 2006 =========

I've not got much further with the FS project, as I've been allocating most of my time recently to learning to program x86 hardware. I've reorganised the memory subsystem of the FS so I can reuse it in other projects as it contains a lot of useful functionality.

I've now started working on many projects simultaneously. Previously I would tend to just work on one project. I will try and restart the FS project quite soon and continue from where I left off.

=========== 19th December 2005 =========

I've now completed all checking of code for the pre-caching filesystem prototype (an intermediate prototype towards a filesystem prototype with full integrated caching). And have begun attempting to compile the new code. I've just completed getting the first problem file past the compiler, it took almost an hour to resolve all errors of this file. Many of the errors were from header files in the wrong sequence.

When I eventually generate the binary, I'm sure it will instantly crash. However it should steadily stabilise, I'm saying this on the basis of what happened with the first prototype. Basically each bug located is an increase in stability. Eventually you run out of bugs, and therefore have stability.

I'll be taking a holiday from computing in the next few days eg Christmas is this weekend. so I think it will be January 2006 that I generate the binary. And really the prototype should stabilise within January 2006, I hope.

=========== General December update =========

I've decided to restart the updates afresh. At the moment I'm working towards the pre-caching prototype. This prototype anticipates an elaborate integrated caching scheme. Because of the complexity of the planned caching I have to get there in phases. The first phase is to get the existing code ready for the scheme. That's what I'm doing now. The coding for the pre-caching prototype is complete. However the checking of the code isnt complete. Once it is complete then the next step is to attempt compiling it.

I've also started learning x86 asm. I've created a new web-page looking at x86 from a 68k-AmigaOS POV. I'm just a beginner with x86 so its an aerial perspective.

Commercial or free?

I havent made any decisions yet but I think I will probably create 2 versions of the project, a free version and a commercial version. The free version should be quite useful in its own right. But the commercial version will be much much better.

As the full project is a huge amount of work it really has to be commercial. In the real world if you want any work done to the place you live you will be charged handsomely for this. Any work at all will usually cost at least £35. 1 day of work will usually cost at least £100. I'm talking about labour costs, you'll have to pay for the materials and removal of rubbish etc in addition.

Now if everybody charges that much for even the most basic of construction work then I think I'm a fool if I dont for my FS work: I have to charge for my FS in order to pay off the queue of people who charge £35 and £100 etc

It would be unfair anyway to other commercial projects if I gave the system for free. I think if someone invests a lot of their private time into a software project that they deserve to make money selling their work. If you sell your work you can then afford to code full time and even recruit people to do the gruntwork + web design + advertising + publicity etc. Basically commerciality allows you to create an "organisation": you use money to organise the activities of others. Even via sponsoring relevant 3rd party projects. So commerciality generates competitiveness by enabling very intense work.

To put it bluntly coding costs money: it costs time and time requires food + electricity + water + ISP charges + phone line rental + central heating charges + other ongoing costs (A). If you try to get coding for free, the money will eventually run out and the coding will stop which is what tends to happen with quality free projects.

Actually I think the trick is to get the price right. If the price is too high you either get piracy or no sales which is what has happened with music CDs. If the price is too low it is exploitation of the author(s): the author is paying (see (A) above) to provide you with the "free" s/w. Most users dont mind paying for s/w, as long as the price is reasonable.

You can only do things for free if you are getting money by some other means or if you restrict the time spent. Eg law/financial firms sometimes do "free" work ("pro bono") for eg charities, but that work is sponsored by the paid work as well as being useful publicity which has monetary worth. :eg it can get media coverage which is free advertising == free money! Commercial companies often give large amounts of money to charity, but this is because they get tax breaks and free advertising (as already explained). "for every £10 you spend we give £1 to XYZ charity" :because that is preferable to giving £1.80 to the government, so in fact it is the customer giving 80p to the company! The government will get their £1.80 anyway via some other route. I would feel happier if they just charged me £9: because if they are forcing me to pay £1 to XYZ charity that isnt charity! charity is a voluntary thing. voluntary contributions are charity, compulsory contributions are tax. charity is organised begging. Insurance is an interesting alternative: tax is fairer though because with insurance everyone pays the same whereas with tax poorer people pay less. Insurance charges you according to your risk factor.

AFAICS most Linux contributors are salaried system administrators and so they are getting the money by other means. We'd all laugh if someone was salaried to admistrate an Amiga! Its too user friendly to require an administrator!

Commerciality is an excellent filter: with free software everything survives regardless of quality and your only guide is a search engine/facility. OTOH a commercial product needs to be good to survive, money rapidly filters in the quality products. go into any high street store: you will see NO free software. For the average high street shopper free software doesnt exist.

the higher the price the more carefully we choose. commerciality is democratic, customers voting with their money.

Really to have a quality software scene you need 3rd party commercial software. That gives the shops something to sell, which finances the advertising that finances the magazines which review the software. It also finances the journalists to research different software scenes, basically doing the research gruntwork for their readers. Information processing is one of the things people buy computer magazines for. Cover disks are also an excellent filter for free software. This information processing draws new coders and new users into the system.

I do believe in free software and open source, but I also believe in commercial software and closed source. I think all 4 concepts are fine depending on the context. eg IMO low level hardware drivers should be open source and OS independent. Ghostscript's inbuilt printer drivers are open source and OS independent: proving that the concept is viable. jpeg is strategically very powerful by being totally portable and open source and almost license free. Different things need to be done in different ways. You will fail if you try to do everything in computing the same way.

It is very naïve to try and make everything free and open: you'll be sabotaged by the commercial companies and their fan bases wherever the free + open work is perceived as a threat. Its very easy to infiltrate an open source project: you merely join and then become an active and terrible contributor. And flame anyone who tries to make any useful suggestions. It is much more difficult to sabotage a commercial project as it is in-house and closed.

I think reimplementation projects have to be free eg Linux, Morphos and AROS. GNU has to be open source and free to succeed: if it were commercial the closed source in house work required would make it too expensive for anyone other than institutions. IIRC that was how GNU got started: it was a reaction to the expensiveness and restrictiveness of commercial Unix software. Unix's problem being that it is on too many hosts.

My FS project isnt a reimplementation and I'm not targetting Unix so I'm not subject to such constraints.

What I'm trying to say is that money is the only reliable reward system. It is tried and tested. Money is the backbone of the economy. You can have free things but the money needs to be rerouted from something that isnt free. eg free forums are generally better than commercial forums, free forums though get their money from advertising and donations. Being "free" the advertisers access a bigger audience.

For the Amiga in particular money is the fuel which creates a dynamic and growing scene. I dont know why but as soon as money is involved you get a big increase in quality. With the Ghostscript project I actually treated the project as if it were commercial as far as that was possible. I thought the program deserved that level of attention. I gradually addressed all the major criticisms that arrived. Recently I created the pamphlet/novel/landscape/billboard/thumbnail schemes because those were facilities I really wanted myself. eg the pamphlet scheme reduces paper and ink usage by a factor of 8. Really these schemes are commercial schemes and in the real world you'd pay for such functionality. I'm not charging as my GS work is partly a learning experience and partly an experiment. (The things I'm saying here are the conclusions of the experiment, namely that quality for free can be done but quality for money will give you 10 x as much quantity and things will happen faster as you can focus full time on the work and even recruit people to help.) If you want to print out a 1000 page document of which 10% will probably be useful but you dont know which 10%, then the pamphlet scheme is the best way to deal with such document bloat and you only need to use 1000/8==125 double sided A4 pages. (when you locate the 10% that is interesting you can reprint that with the novel scheme: full a4 size is usually unnecessarily big). If 500 pages costs £2.60, then you've saved (1000-125) * 2.60/500 ==£4.55 in paper. And (1000-2 x 125) pages of ink, if 1 page costs 5p of ink then you've saved (750 x 5p)==£37.50 of ink. So you've saved £42.05 by using the pamphlet scheme. Its as if I gave you £42.05 for using the pamphlet scheme! So from the POV of money saved it would be fair to charge £20 just for that facility.

in the real world one often spends money in order to save money.

Money is what makes x86-windows so dynamic. From the software POV the Amiga in its hey day was even more dynamic, IMO because they got the balance exactly right. Newcomers to the Amiga scene are probably unaware that Electronic Arts began as an Amiga software house producing endless classic titles for the Amiga 500. They made full use of the cheerful 4 channel 8-bit sound and 32/4096 colours when other computers had 16/16 colours and either no sound or just beeps. Even today the Amiga is the only computer system which is enjoyable to use. Today EA have left the Amiga and advertise 3D games on national TV stations. Anyway their big presence is based on the fact that they were always commercial and they've ploughed back the money they've made wisely. eg TV adverts cost big money, and clearly they route some of their profits into advertising.

Money does have a certain magic, where money flows everything glows. Its why people become addicted to shopping. This is why Windows glows and Linux lurks. FACT: Linux's future will never match the Amiga's past: simply because the Amiga was blatantly commercial and Linux is blatantly "free". eg lets say in 1990 the Amiga dominated the high street: Debenham's (via Silica shop), Dixons, John Lewis, all sold the Amiga. 80% of computer games shops eg Virgin games were Amiga games, Amiga magazines dominated the shelves: CU Amiga, Amiga Format, Amiga Shopper, Amiga User International, Amiga Computing. The Amiga's presence was 10 x the presence of MacDonalds. Amiga programs especially games had a bigger presence on the high street then than XBox and Nintendo and Sony can ever dream to have. Thats because the current companies are too centralised, the Amiga had a ginormous 3rd party commercial scene which was totally decentralised. This is why I said earlier that you need an active 3rd party commercial scene, AFAIK the Amiga's past scene has never been outdone by anyone. Money really is what makes the world spin, if you took the 50 best programs on the Amiga public domain and made them commercial a new scene would start to grow.

Think about free programs you use, now think what price you'd feel happy paying for said programs: eg with 68k-gcc I'd say I'm prepared to pay £80 for it. That means everyone using it for free is robbing the authors of £80. As gcc is very highly portable there is an economy of scale so maybe from that POV it is worth say £40. If you look at your free programs and decide on a perceived value for each program such as the £80 or the £40, then the authors are being deprived of all that money. If that money went to the authors they'd go the extra mile and develop full time as well as refine and expand their work. As well as finance a support framework: help-site, tutorials, related utilities etc. We've probably each got maybe £500 perceived worth of free programs, if there are say 2000 users in the scene that is £1000000 which is missing! Had that money been channelled to the developers it could have led to 100 classic titles (thinking of £10000 per classic title). Those 100 titles would draw in new users boosting the user base by a factor of 10 to say 20000: you are moving towards a sub-economy. 10% of the user base can actually make some sort of career from the cash flow.

So in fact free software is a lose-lose situation.

Commodore-Amiga were good at making sure there was a suite of quality commercial programs at the time of releasing new hardware. They didnt bundle software but co-operated with 3rd party s/w houses to synchronise the release of the new s/w with the release of the new h/w. eg there were probably at least 50 apps/games available when you bought an A500. Including many by Electronic Arts. I think there were also at least 50 apps for their CD32 machine, including IIRC a teach yourself Japanese CD! Escom's error was to bundle s/w with the machine, as a result buyers were less inclined to purchase. Bundling also stifles competition, eg if you bundle an art program it removes customers from competitor art programs. Those competitors then leave the scene, and the bundled prog's company also leaves the scene as they get less money from a bundle. My XP machine was bundled with about 50 titles, of which I've probably used maybe 5 (browser, email (outlook express), CD-burner, CD-player, ???) and I've only bought 1 title, and used 0 games. I prefer YAM for email, and only use Outlook Express when YAM gets stuck. I only use XP programs when I cannot do something via the Amiga as XP progs are very unergonomic (though pretty). OTOH I've bought a lot of peripheral h/w all of which I can use from WinUAE. IMO h/w bundling is always a bad idea, as they'll bundle all the stuff that isnt selling. They'll also bundle substandard or unnecessary h/w, eg they bundled a titchy hard disk and a DVD-ROM (I want 2 DVD-writers!).

Where today people set up collective open source free projects via the internet, in the old days people would set up collective commercial projects especially games without the internet. You'd go to an Amiga show and there'd be 10 people of some games s/w house wearing the same T-shirt! (10 identical T-shirts that is!) And they'd be generating a new game each month. They'd have a customisable games engine, and would make new games mainly by reparametrizing. Those were the days!

A really interesting game was Boulderdash where (in some versions at least) the user could create their own screens. The game was basically a run time games engine with 100? parametrization files one for each screen! The amazing thing is that generally speaking each screen involved a totally different concept. This was clever because they'd factored out programming-skills from games-design. IMO Boulderdash was one of the best games ever, each screen of Boulderdash outdid many other games. (it may depend on which implementation of Boulderdash you try).

AMOS was another unusual concept where they seemed to have enhanced BASIC so that commercial quality games could be created with BASIC. If you werent told you'd think an AMOS game was written in asm. In a way AMOS was a games engine BASIC. If AMOS interests you there is a huge amount of AMOS material on coverdisks and magazines of the past. I think some magazine even gave a version of AMOS for "free" and then did tutorials on how to use their free version. (The Amiga Format collection cover disk no. 10 was the full version of AMOS. My A1200 isnt connected up at the moment, as soon as it is I will determine when this cover disk was.)

In general I think compiled BASIC can be almost as fast as asm and faster than C. But only if someone puts in the effort to create an efficient compiler. A big incentive for doing an efficient BASIC compiler is it has a huge potential market. I could create one myself but I would have to commit the time and generally speaking I avoid reimplementation projects. I use reimplementations (eg Geekgadgets) and emulators (eg WinUAE) but I dont do reimplemention or emulation coding. I prefer the stratospheric excitement of creating things which are completely new. Another rule I follow is to only work on projects I want to use myself. Also I'd have to implement all the things users expect which will cost further months of time. Such a project would have to be commercial.

Maybe I should reimplement Arexx instead but IBM probably own the ©opyright (via rexx?). And I'd have to learn Arexx first in order to reimplement it. I know enough about Arexx to know I can reimplement it. Arexx anyway should be fully available via Amiga Forever, so its non urgent. Its not like I'm not busy! My computing to-do list is a mile long. IIRC OS4 cannot bundle Arexx for ©opyright reasons, I dont know if that is just the ©opyright for the binary??

AROS and Morphos are healthy projects with steady currents of development and both are subsidised. Although the binaries are free the OS development does get money: Morphos development is subsidised by Pegasos sales. AROS development is funded by their bounty system, users volunteer money for OS facilities they want and developers receive that money on completion. I am at a distance from both AROS + Morphos so I dont know the precise mechanics of the funding. Hyperion are completely beyond my field of view so I cannot say anything at all about them.

In the Amiga scene the Pegasos has had the biggest presence entirely because it has the biggest flow of money. Linux despite being free is used by a lot of big companies eg I emailed someone from uni who is now on the staff at a different university and the email header showed that the latter university uses Linux. I think IBM and the german government use Linux, so there is commercial pressure for Linux development. Essentially Linux development is funded indirectly by the companies that use Linux. A lot of manufacturing companies seem to use gcc for their embedded chips but its a complete mystery to me how gcc is developed.

Disclaimer: please note I am not endorsing AROS or Morphos, but am just observing matters of fact. All I can endorse are the Amiga 1200 and OS3.9, as well as Commodore-Amiga and Escom.

Note that even with free education and free libraries, all the staff are salaried! Most charity workers also are salaried (not a big salary), priests also are salaried through donations. It is just with computer programs that people expect major amounts of expert work for free. See above why free software is bad for everyone, bad for developers and bad for the users.

Computing is a battlefield of ideas, each project is based on a suite of guiding principles or ideas (not always good). Today the battle is mainly on the internet. With the A500 the battle was mainly in the magazines and through direct contact (my Amiga is better than your Archimedes/ST/Mac/PC). The scene was 100 x as healthy then, you can time travel to any point in the Amiga's past via magazines. Eg to revisit 1992 just locate an Amiga mag from then. Its interesting to read journals from the past because the journalists will be second guessing the future whereas you are from the future so you know what actually will happen!

A given project will usually have good guiding ideas and bad ones. Each project has a different emphasis. Often new projects are started in reaction to the bad ideas of existing ones. And usually such new projects introduce a new set of bad ideas. It is very complicated to get things completely right.

I intend to continue creating free software but it will be subsidised by things which arent free. With any commercial product I create I wont be intimidated to make the product free just because a competing product is free. If I evaluate their product as being worth £20 then I evaluate that they are giving away £20 each time someone downloads their product, so I'll probably email them and ask them if they can send me the £20 directly in cash!

I will charge a price that I would regard as a fair price were I a customer. The product will be part of the economy and will follow exactly the same rules that apply to any item in any shop whether it be a fruit or a lorry (toy or real or a photo!).

Comparison to other projects

There are various projects where someone out there has decided there is a better way of doing things than the official AmigaOS system:

Turboprint:

here the author realised something much better could be done than AmigaOS's printer subsystem. The official system could only do 4096 colours via 12 interleaved bitplanes. Also the rendering quality subject to that constraint was bad. Turboprint's author created a system that replaces the official system, which supports a wide range of modern printers, supports the full truecolour range, supercedes the interleaved bitplane approach via various modern options eg byte per component, has fast rendering algorithms, has high quality rendering: the image quality + colour quality is essentially perfect. The work is very perfectionist. Essentially AmigaOS's scheme was amateurish whereas Turboprint gives commercial quality printouts.

Cybergraphics:

here AmigaOS's interleaved bitplane + palette number scheme is superceded by the various modern approaches eg byte per component, 32-bit word per pixel, 16-bit word per pixel (5:5:5 and 5:6:5 schemes) and with a modernised API. It was basically a full modernisation of AmigaOS's scheme. Interleaved schemes are really dreadful: complicated + slow, a lot of work to process a pixel. pixel or component arrays are much faster and simpler to code eg you can process either a pixel or a component in 1 asm op depending on which scheme. Interleaved is totally obsolete. If the A500 had used a pixel array scheme it would have been even more impressive than it was: eg they could have used a 16 colour array of 2 pixels per byte scheme or something. I think AmigaOS realised about pixel schemes with their CD32 machine (cousin of the A1200) which used pixel arrays of byte per pixel.

I encountered the above 2 projects via my GS8 port. At the start of the port I didnt know about them, I learnt about them from user feedback.

MUI:

AmigaOS originally did windowing via intuition.library, then they attempted some object oriented schemes eg Boopsi. With MUI someone out there has done their own object oriented windowing scheme. Windowing is probably the ultimate place for the object paradigm. The "magic" is the magic of object oriented redirection, object orientation is magic because all sorts of interesting things happen automatically through cascades of redirection. I've never programmed with MUI myself, but I use so many programs which use MUI that its clear that MUI totally outdid AmigaOS's schemes. I have tried coding BOOPSI but its really complicated to do even something very simple, and I think it is a bit nonstandard.

SFS:

here someone has looked at the limitations of the official scheme and forced through a better system compliant with the official scheme. I am of course a competitor to SFS, that doesnt stop me acknowledging SFS's merits. SFS is certainly a successful project and AFAICS has achieved the objectives it set out to achieve.

scsi.device:

here Commodore-Amiga co-operated with 3rd party SCSI h/w companies to create a sensible scheme that could be used by any 3rd party company. The action was by the 3rd parties and Commodore coordinated this. Commodore could and did co-operate with 3rd parties, but only the big players. SCSI is a cross platform API which needs an extension of the exec i/o API, #?scsi.device achieves the required API extension: it allows you to control hardware invented after your SCSI interface was manufactured. eg I've programmed SCSI CD-RW which IIRC is SCSI-3 through a SCSI-2 interface via scsi.device.

I wont mention AROS + Morphos + OS4 + WinUAE as those are warzones,

No company can monopolise innovation, neither AmigaOS nor Microsoft can. Microsoft are different from AmigaOS in that they know how to recognise external quality projects and they then incorporate them into Windows.

Unless you are very senior it is very difficult to innovate within a company framework. Junior employees who attempt to innovate are regarded as a threat. So the best innovation tends to be from hobbyists. High ups in a company dont like being outdone by lower downs, there is a pyramid of egos.

My FS project differs from the above projects in that I'm not interested in backwards compatibility with the official system. I will try and achieve compatibility with AmigaDOS and vice versa but only afterwards, AmigaDOS compatibility will reduce the power of the system. My project has no dependencies on AmigaOS, but to run it on AmigaOS the required layering brings in exec.library and i/o device dependencies. If I port my system to Windows or Linux then any program that depends on my system will also port. The portability is much cleaner than GNU portability as GNU portability tends to be *nix-portability and isnt general portability. Linux portability is even worse and is linux-portability via (IMO deliberate) Linux dependencies.

A really powerful feature of my project is that it is well suited for an OS bootstrap. Later in the project I will try and improve the bootstrap functionality.

The objectives of my FS project are possibly over-ambitious, that doesnt matter: if necessary I can restrict the objectives later, currently anyway the objectives are being done in phases, each phase being quite realistic. I have no idea how long it will take even to achieve the immediate objectives.

Also I dont know how it will compare with other projects such as SFS. No idea at all if it will be faster, I certainly want it to be faster but maybe it wont be. The project is incomparably different to SFS, my project doesnt need to be faster than SFS to succeed, as speed is not the only thing I'm looking at. The Acorn Archimedes was 4x as fast as the A500 but I'm glad I bought an A500. The Amiga's OS was light years ahead of the unitasking Acorn Archimedes and the Atari ST. IIRC later on the Archimedes acquired co-operative multitasking, but it was never to be in the same league. I am certainly looking at speed but other things are just as important to me.

In fact I dont look at SFS at all with my work, the only comparisons so far done have been with NTFS as that is the central MS XP FS. Nowadays I use the FAT FS's instead of NTFS as they can be accessed from XP + Linux + WinUAE. NTFS partitions dont show up on the Linux I use, unfortunately most of my XP partitions are NTFS. I dont even look at AmigaDOS, I only mention SFS because that is what many people mention when I mention my project. I dont look at SFS because it isnt addressing the same questions that I am addressing so it wont help me with my project.

I dont want to compete with SFS by copying them, that isnt sport. Actually I'd been thinking about this project years before I heard about SFS, I thought about doing a FS when I created my CD-RW programs approx. 2001 and heard about SFS in 2003. Furthermore even after hearing about SFS in 2003 I had other projects ahead of my FS project in my to-do list. If my aim was the same as SFS's aim then probably I wouldnt bother doing this project.

if you mention the Amiga a lot of people then reply Linux or Windows, so if you talk about the Amiga you have to mention Linux and Windows eg you can say that AmigaOS is much more asynchronous, in particular AmigaOS windowing is maximally asynchronous. In many scenarios XP's windows become totally synchronous, where only one window can be accessed. Linux is even worse where the entire OS will freeze up if you attempt certain asynchronous ops.

Now because everyone insists on mentioning SFS I will compare performance with SFS, however not till much later on. I will also compare performance with other projects such as NTFS and Linux's FS's.