Vampire discourse, keep it civil (was: Vampire 1200 V2 waiting times)

[TCD]

This thread is such a train wreck my old self would actually feel sad about it. But you guys really do a great job of reminding us why 'Amiga' never had any future beyond CBM's bankruptcy. Keep at it boys!

[daxb]

But first you need to revive Amiga. ;-)

[TCD]

Originally Posted by daxb

But first you need to revive Amiga. ;-)

Apparently it never died

[phx]

Originally Posted by Korodny

How is this worse than not having any forks after "the development comes to a halt" because the source code is not available legally?

Judging that also depends on your personal expectations. Do you want further development of AmigaOS at all cost? Then open source may indeed be the only chance. But once the first open source fork of the former "real" AmigaOS is established, the box of pandora is open. Many will follow and exist in parallel, as there is no real owner who cares anymore.

I guess that our expectations differ in this point, because I am talking about AmigaOS/m68k for classic Amigas only. New architectures have already proven to be a failure in the past. What big improvements would you want to add there? The OS will never become usable by modern standards again. Maybe we can fix the most important bugs and have fun with our hobby. But that's it.

AmigaOS not being free software has not stopped the 'forks' of the mid-nineties (AROS, pOS), it didn't stop MorphOS from becoming a reality

Maybe we should define "fork". For me a fork is based on the original source, which is not true for any of those. They are completely different operating systems for different architectures (x86, PPC). If I wanted to release a program for AmigaOS/m68k then there was still only one OS.

and it has not stopped A-EON from starting "System v54" recently.

Interesting. Never heard of v54 before. A PPC OS? So also irrelevant.

most of these projects are closed source, so everybody has to keep reinventing the wheel.

Fortunately. This keeps the number of variations small. And many will disappear again.

I know, which is why I actually answered. I was puzzled to hear this from you, of all people.

Might be because we have different expectations for the Amiga's future. See above.

"one defined target" - like this? Ah they joys of a closed source eco system...

I could have also included MacOS and Linux ports. I even did SGI Irix ports of it. But there is only one AmigaOS/m68k port, which runs on all classic 68k Amigas.

[Korodny]

Originally Posted by phx

because I am talking about AmigaOS/m68k for classic Amigas only. [,,] What big improvements would you want to add there?

Improvements to the core of m68k-amigaos or changes to the API? None, obviously - since backwards compatibility would have extreme priority on that platform. But that would be true for anybody trying to create a fork, I don't see how an open source AmigaOS would negatively affect that:

OS3 application developers - i.e. all 1.5 of them - have been targeting 3.1 + P96 + MUI for ages. Games developers basically stick with naked 3.1 (or even 1.3). Every OS developer would be aware of that and keep these APIs untouched - or disapperar into obscurity pretty fast. But that doesn't mean 3.5/3.9 or 3.1.4/3.2 didn't bring lots of interesting and useful enhancements. Maybe not so much for developers, but users love(d) these releases.

And this thread is (was ) about the Vampire, which has a lot of fans - Gunnar recently claimed 10,000 units were sold. You might not want AmigaOS development to continue, but these customers certainly do.

Maybe we should define "fork". For me a fork is based on the original source, which is not true for any of those.

That would still leave three m68k forks (the H&P branch, the Cloanto branch, the Hyperion branch). And patches/replacement libraries from third parties. And CoffinOS.

But I find this sudden emphasis on "original code" unconvincing. History suggests that branding has much more effect on Amiga users than "original code". People bought the stupid AmigaDE SDK en masse, and Ben Hermans is taking big risks to maybe get his hand on some of the trademarks - despite already having (or so he claims) "the original code".

[Bruce Abbott]

Originally Posted by Thomas Richter

Forget the "We" for the moment. Be honest, and say "I", because it's certainly not "We". And no, "autoconf" does not take ages to probe, that's just nonsense.

Where did I say it did?

I'm so proud of you.

No you aren't. You are combative and full of snark. So much for 'keep it civil'.

CBMs expansion cards surely did autoconf. Their motherboard hardware did not, but CBM was CBM and CBM was cheap.

Yes, CBM was cheap - and I'm glad they were. If CBM hadn't been cheap the Amiga would have been overpriced and unpopular. To be popular it had to be cheap, just like all those other home computers of the day. Even PC clones. They didn't get autoconfig until 1995. Dealing with hardware conflicts was a nightmare, but people bought them in droves because they were cheap. Well not all clones. Commodore PCs had a form of autoconfig, and they were well made and reliable too. But people didn't want that, they wanted cheap.

Frankly, you have no clue about software or hardware design in an open system....

For someone who insists that the Amiga is a 'toy' you seem to be taking it way too seriously.

"I'm too lazy to learn, so I ignore" seems to be a good rationale to design a system indeed.

I am 65 years old. I don't have the time or energy to fill my head with stuff about other systems that I won't be working on. I have collected a number of other home computers with the idea of learning about them too. But I have to face the fact that I don't have an infinite amount of time left in this world, so I need to decide which (if any) platforms to keep - apart from the Amiga of course.

Actually, the FPU is not very helpful to display 80 bit IEEE, you need software for that in first place.

Obviously. Can you direct me to any example code that doesn't use floating point?

But Motorola, other than our hero, was wise enough to provide a mechanism that such software would continue to run, albeit slower, but still providing (for integer) identical results - and for floating point, results of comparable precision. That is the difference between a design by engineering, and the design from your gut. It's simply not wise.

I'm confused. What important features did they remove from the Vampire and then not provide software emulation for?

Ah, so you "applaud" without knowing what to applaud to?

Yep. I applaud diversity and innovation, even when I don't know its intimate details.

It is not a MMU, actually. It has two different page sizes, and only two "mode bits" for each page size that will become too limited for a DMA based system anyhow. It is not very forward looking.

You do know that most Amigas don't have an MMU, right? Even though they are all 'DMA based systems'. Terribly limiting and not very forward looking I know...

[Bruce Abbott]

Originally Posted by Thomas Richter

Right, and we would have about a dozen slightly but not totally compatible versions, and endless discussions in forums which is the right and proper one. Why that is potentially helpful to have more incompatibilities is beyond me.

I agree. We need a 'gatekeeper' for Amiga OS, preferably one with a commercial interest so they have to stand behind their product.

Unfortunately we had two 'gatekeepers', each of whom thought they alone had the legal rights to it. As a result the Apollo Team took the only legally viable route - AROS. A shame, but not their fault. Such a pity that people can't get along...

My personal take on this is that on a (non-)community as acid and unprofessional as this one, it is not particularly helpful to open source stuff. There is no benefit except endless discussions, and it would certainly not improve the quality of the (already lousy) code.

I agree. The OS is supposed to be a 'black box' and if well coded and documented we shouldn't have to peek at or modify it.

OTOH well written source code can be educational to read and may help with understanding how your code interacts with it. If I (ahem) had the source code to eg. Amiga OS 3.1 I wouldn't consider trying to change it in any way, but I might inspect it to help make my own code better, or just for general entertainment.

[Bruce Abbott]

Originally Posted by Gorf

Now 3.2 is redoing stuff that was done in 3.9 many years ago .. and if Hyperion goes down (which it almost did a couple of times already) someone else will have to start again from 3.1 sources...

3.1.4 did a much needed reset back to 3.1, but there is no going back now. As time goes on I expect Amiga OS will become more bloated and less useful to those who wish to support classic Amigas. In my own code I try to support OS 1.x as much as possible. Whether I ever upgrade myself - we will see. Maybe with version 3.3.

This has not much not do with Vampire except Gunnar wants to "free" AmigaOS from copyright lawsuits ... which is kind of a mystery to me.

What's mysterious about it? The legal hazard was clear, and when you have a lot of hardware you might not be able to sell... that was the proximate reason for Commodore's collapse.

ApolloOS with AROS as foundation might be the right way: bringing it on par with legacy 3.x in terms of speed and compatibility, would finally make Hyperion and Cloanto obsolete, which would be a good thing.

Not holding my breath....

[Cego]

Interesting. Who says that the involved parties have any claim over copyright/license at all? Maybe none of them have the right to develop OS3.1?
Just because somebody claims to have the legal basis to do stuff, doesnt justify your actions. AmigaOS3.2 - as of now - is also a pirated OS3.1. And you Thomas, are no better than Gunnar in this regard. You worked on it and contributed knowing that the legal issues are not settled. You even justified the development by saying that either it's now or not at all. Lets do it while we can...

Its hypocrisy at its finest to accuse someone of doing illegal stuff, while you willingly agreed to work on an OS development that has been illegal.
Right now Hyperion and a lot of Amiga shops are selling illegal copies of AmigaOS3.2. Also a lot of developers are working on it illegaly right now.

Regarding FPU and Autoconf: I dont hear you mock about LC/EC versions of CPUs or accelerator/ram expansions that have to be intergrated via addmem and kickstart modifications (like TF1260 IDE). Why does the Vampire bother you so much? I would also welcome a clearer approach of integrating the Vampire into the system, but this is peanuts in the end and not worth such a constant shitstorm from your side over the past years.
Its really tiresome to witness that you're still on your personal vendetta against Gunnar.

[Gorf]

Originally Posted by Bruce Abbott

What's mysterious about it?

How he is going to archive that.

[OlafSch]

Originally Posted by phx

I feel I shouldn't post in this thread, because I have no opinion concerning the Vampire.

But I want to add that Thomas is right about not open-sourcing the OS. The only possible approach would be to have a group of well-respected Amiga developers to finally decide about the inclusion of new patches. But that's more or less what we have now, as the source was already leaked long ago.

An uncontrolled open source AmigaOS would be doomed. Neither a user nor a developer should ever want that, because forks will happen, making the situation worse with every fork. As a developer, which one will I support? Does my program crash on other forks? Do I have to test all of them? Nightmare! It would harm future developments extremely!

yes and no. The biggest problem occurs when developers start to mix code from different sources and replacing parts that were not created for each other. In this case I mean replacing aros components (f.e. libraries) and add patches to improve perhaps 3.X compatiblity or speed something up and use patches from 3.X world. That can certainly make a system unstable (I experienced that even with a commodity in my distribution that created suddenly problems that seemed not related) and makes it difficult to find the reason for problems. I stay "original" where possible and only add stuff. Also it becomes a problem if people fork and do not give back changes and expecially change the API making it incompatible.


@Korodny

I use the official Aros branch that is currently cared and updated (Deadwood) for my distribution and the same sources are used for other platforms like X86. Of course I add stuff.

On the other hand we already have different systems, next to 3.1, 3.5 and 3.9 we now have 3.1.4 and 3.2 and almost every system is more or less heavily patched. Even P96 is a kind of patch. That is also the case for many distributions like amikit. So where and what do you test? Plain 3.1?

"Workbench, Ambient, Directory Opus Magellan, Scalos, Wanderer."

these are desktops and should not affect software. You mix API (OS) and desktop (GUI) now

[OlafSch]

Originally Posted by phx

This may work, with a strong company behind AmigaOS.

But let's assume there are legal issues and the state is unclear, or just that the owning company is "weak". Maybe development comes to a halt. Then a fork of the whole source tree can be done and ambitious developers might bring AmigaOS to a new level.

Sounds good at first. But this is already the first fork. How many users or developers will stick to the original OS nevertheless? And after some time, development of this fork slows down and eventually stops as well. No problem, there will be the next fork, with other developers who think they are doing the only real AmigaOS. And then another one, and another one...

I am open source developer at the NetBSD foundation (was the OpenBSD fork good?) and I release everything as open source myself. But for AmigaOS I would really prefer when it stays closed. Even if it would mean that development stops. But then at least you have one defined target to write your software for.

but if I take you seriously there should not have been 3.5 and 3.9 and expecially no 3.1.4 or 3.2 update either. Not everyone is willing to update, be it because happy with existing system or fearing compatibility problems or not wanting to give money to a certain person so you have a new "fork" there either. Fork in sense of something potentially incompatible you would have to test. Many use individual installations with specific combination of patches and addons. Hardly anyone uses plain 3.1.

A comment I just found that summarizes it:
http://eab.abime.net/showthread.php?...os#post1557189

"Well, it sounds like a lot of work to do. Much respect for your efforts. Unfortunately the AmigaOS is all but easy, and only very few people are able to handle it. I did try to put mu hands and learn, but then, i have found that experience is what is really needed, cause to obtain good results, you need to use alot of 3th party little applications and patches, that you must know. And there are many version of them to do exactly the same thing. Only someone who is playing with Amiga since 90s is able to fully understand what to do and how, with witch applications and so on. So, yes, only thanks to people like you, who makes efforts to prepare something good and ready, it is possible to enjoy Amiga as it's best."

[Thomas Richter]

Originally Posted by Cego

IJust because somebody claims to have the legal basis to do stuff, doesnt justify your actions. AmigaOS3.2 - as of now - is also a pirated OS3.1.

Cut it short please. On which basis is 3.2 "illegal"? As you probably know, Hyperion does have a development licence for the Os. Gunnar does not have that, leave alone a right to distribute copies.

Originally Posted by Cego

And you Thomas, are no better than Gunnar in this regard. You worked on it and contributed knowing that the legal issues are not settled.

There is a subtle difference between "contributing to company that is in a court action" and "just taking the binaries and redistributing them", right? There is a subtle difference between that and "just ignoring written will of developers that P96 drivers need to be paid for". And yes, you can clearly read this in the 2.0 version of P96 in Aminet. Go there, download, read.



It's this type of action that drives me mad - just ignoring people's expressed intent.



Well, if you do not want to license AmigaOs, fine, do without, go AROS. That's fair. What's not fair is neither licensing it, but just spreading pirate copies.


Well, P96 authors don't want to license to you? Well, bad luck, consider alternatives. But just taking the driver package and ignoring them is neither a "friendly action".



That's not how you develop a hardware platform.

Originally Posted by Cego

Its hypocrisy at its finest to accuse someone of doing illegal stuff, while you willingly agreed to work on an OS development that has been illegal.

Stop the nonsense please. On which basis is 3.2 illegal? That I really want to understand.

Originally Posted by Cego

Regarding FPU and Autoconf: I dont hear you mock about LC/EC versions of CPUs or accelerator/ram expansions that have to be intergrated via addmem and kickstart modifications (like TF1260 IDE).

Which are bad pieces of hardware, yes, indeed, and I'm not advocating to use them, but they are only that. Their makers do not come along and tell me that this is the one and only system and future.

Originally Posted by Cego

Why does the Vampire bother you so much?

Because there is also a subtle difference between "building a turbo board" and "claiming to reinvent the Amiga".

Originally Posted by Cego

I would also welcome a clearer approach of integrating the Vampire into the system, but this is peanuts in the end and not worth such a constant shitstorm from your side over the past years.

It would be peanuts if its maker wouldn't claim that ignoring system specs is the future of the system.

[Thomas Richter]

Originally Posted by Bruce Abbott

Where did I say it did?

Here you did, that was yesterday:

Originally Posted by Bruce Abbott

We don't want to be like PCs that take ages to boot up while they probe hundreds of different hardware devices that probably aren't present (and don't always get it right). Autoconfig imposes a standard way for plug-in cards to tell the OS what they are and get the resources they need.

Autoconf is *not* like that. It is a relatively simple protocol if you implement it in hardware. If you just push a ConfigDev structure into expansionbase to advocate your hardware you detect yourself, it's even simpler.


The really big advantage is that you do not need "arcane byte poking" to detect what hardware you're talking to. There would be a standard way how to detect a vampire or its RTG graphics, for example, namely by looking up its vendor and product ID.

Originally Posted by Bruce Abbott

Obviously. Can you direct me to any example code that doesn't use floating point?

For converting IEEE 80 bit float to ASCII without an FPU? Sure, here you go:



http://aminet.net/dev/lib/ThorLib.lha

Originally Posted by Bruce Abbott

I'm confused. What important features did they remove from the Vampire and then not provide software emulation for?

Removed? There was nothing "removed", but it would have been possible to create an emulator trap for the FPU for example, for those that need high precision, and fix that by software. I guess I already explained that.


I understand the limitations of the hardware

Originally Posted by Bruce Abbott

You do know that most Amigas don't have an MMU, right?

That depends on your definition of MOST. Do you have a statistics? I don't. All I can say is that those have a 68040 or 68060 family member do. They have to, getting DMA right on a processor with copyback cache is a problem. Getting the EC CPUs working in the system is creating a headache from the system architecture pov.

Originally Posted by Bruce Abbott

Even though they are all 'DMA based systems'. Terribly limiting and not very forward looking I know...

Not really, and again, you're talking about things you probably do not fully understand. There is no problem if you have writethrough only. The trouble begins with copyback, and thus with the more advanced members.

[daxb]

Originally Posted by Korodny

And this thread is (was ) about the Vampire, which has a lot of fans - Gunnar recently claimed 10,000 units were sold.

But you have to interpret the 10.000 units. That means at best 5.000 users maybe less. A lot of users have bought two/three units and then there are collectors and some greedy profit people who speculate.

That would still leave three m68k forks (the H&P branch, the Cloanto branch, the Hyperion branch). And patches/replacement libraries from third parties. And CoffinOS.

You seem to have a special definition of what is a fork. OS3.5/9 and 3.14/3.2 are just versions. Cloanto 3.X is a distribution of collected software/patches. Nearly zero "real" development was done. So phx is right with only one 68k OS if you don't take AROS 68k (known as ApolloOS "nobody" is using for a reason).

Unfortunately we had two 'gatekeepers', each of whom thought they alone had the legal rights to it. As a result the Apollo Team took the only legally viable route - AROS. A shame, but not their fault. Such a pity that people can't get along...
.

Maybe you don't know or ignore the facts. First, Apollo Team used illegal Coffin distribution. Next for some low number Vampire units a license from Cloanto because the pressure from the "community" not using Coffin was massive. Then the switch to AROS because it was free. The Apollo Team had and still has the choice. The suggestion to deliver a install script/instruction that makes it possible for the user to install OS3 was denied. So there are three legal ways to get an OS on the Vampire system.

[OlafSch]

@daxb

I think "fork" is here not used correctly, more precise would be "not so compatible versions" or how you would call it. Most people with individual configurations are heavily patched, all distributions on 3.X are heavily patched, then there are the older 3.5 and 3.9, all adding new stuff or at least different stuff, then 3.1.4 and 3.2. Not everybody certainly has changed to it. So you have lots of different configurations. What is your base for testing? The problem ist not new. In the old days with Commodore and a commercial market new OS versions were less problematic because there was a big company behind it and developers adapting the software (if needed). Today it is different.

Regarding Aros... do not use it if you do not want it. Simple as that. I know Aros 68k quiet well so I know what I write about...

[grond]

Now that we discuss Kickstart/AmigaOS version after 3.1, there is an interesting parallel to AMMX: would anyone seriously write a program that does not run on 3.1 but only on 3.1.4 or 3.2? I doubt it. If there were such programs, how would that not be scandalous vendor-lockin?

[OlafSch]

Originally Posted by grond

Now that we discuss Kickstart/AmigaOS version after 3.1, there is an interesting parallel to AMMX: would anyone seriously write a program that does not run on 3.1 but only on 3.1.4 or 3.2? I doubt it. If there were such programs, how would that not be scandalous vendor-lockin?

I somewhere read that stuff from 3.1 not worked on 3.1.4/3.2

if you change something there is always the risk of something broken. If you want to avoid that you should only use plain 3.1 (or 1.3), of course without any patches, newer libraries and so on and of course no 3rd party software like P96. How useful this is, is another question.

[Thomas Richter]

Originally Posted by grond

Now that we discuss Kickstart/AmigaOS version after 3.1, there is an interesting parallel to AMMX: would anyone seriously write a program that does not run on 3.1 but only on 3.1.4 or 3.2? I doubt it. If there were such programs, how would that not be scandalous vendor-lockin?

Indeed there is, and I'm not that much of a friend of creating too much of additional interfaces there. 3.1.4 was really just an update of 3.1 without much additional logic, and it is pretty much the same for 3.2. There are features users get "for free" because they do not depend on such extended interfaces.


The same goes for P96 as well: The current additional features you get do not require "exposure" to the user space, so to say, i.e. there is no extended API. All programs can profit from them, and there is no need to update your all-day software to take advantage of, say, screen dragging or multi-monitor support. That works all "out of the box". At driver level, screen dragging will require additional functions, but it does not break old drivers. Multi-monitor requires even nothing to be done at driver level, it just works from the core.



There were requests in the past to add functions like "blit through an alpha channel mask in true color" to the P96Api.library, but so far I'm not that much a friend of such functions because they would all stay quite isolated, without much software that could actually use them.


I'm not saying that they probably don't get reality at some point, but that clearly does not have priority. Priority is "get the thing working as smoothly and as compatible as possible".


I think exactly the same about the vector extensions: They don't do harm, but I wouldn't really need them. What I'm more worried is that there are compatibility issues that have not been solved, and instead a relatively useless extension has been added to the core.



Thus, I'm not against them, but I do not see too much benefit in them. There are other things that I consider much more important than this.


Getting the balance right here is not so easy indeed.

Edit: There are, of course, cases were extending the library API makes a lot of sense because the Os itself uses the extended API. To give you two examples: The V45(!) rexxsyslib.library has extended API interfaces to retrieve the values and set the values of REXX variables. This RVI is/was part of the amiga.lib. Now it (also) sits in the rexxsyslib.library. The reason is that multiple Os programs (e.g. C:Ed) depend on the RVI functions, and by offloading them from amiga.lib to rexxsyslib.library, I was able to squeeze out a couple of KByte from the distribution disks to make the result fit on disk. Another example is the 64-bit division in the utility.library. A couple functions depend on such functionality, e.g. HDToolBox, as it is required there for large disk support. It certainly makes sense to offload such functions to utility for Os tools to use them, as it makes simply the code more compact. GadTools scaling and font sensitivity is a third example were all the Os prefeference editors work with, and even if no single user code will ever call or use this functionality, it is a win because the Os programs themselves are now shorter (or at least not longer) and deliver additional functionality.

Thus, it is not *quite* in parallel with extended instructions, but there is some parallelism indeed.

[jbenam]

Originally Posted by daxb

But you have to interpret the 10.000 units. That means at best 5.000 users maybe less. A lot of users have bought two/three units and then there are collectors and some greedy profit people who speculate.

Gunnar has an agenda. He needs people to believe in his little cult and his vision of a "reborn Amiga", otherwise the Vampire won't sell. And this is fine and dandy, he has his little army of cultists repeating that the Vampire is the only way forward and that everyone else is EVIL and wants to keep Amiga dead for their nefarious purposes.

What about the rest of the Amiga community that didn't drink the cult Kool-aid? He needs exclusive software to convince these users to purchase one, but developers won't develop for a board with a small number of units sold. Gunnar also isn't an honest person (and this has been demonstrated numerous times) as he doesn't care about licenses and/or pirating software so he wouldn't have any qualms with lying on the number of Vampires sold.

Sooo... do I believe that 10.000 Vampires were really sold? Only if there was a video of them all laying down on a table