OpenSource for Amiga.

[xeron]

bloodline wrote:
Before we need any Opensource Apps, we need an Open Source Operating system  :-D

Wouldn't you agree?

In all honesty, absolutely not. This is not a sleight against AROS, which I'm interested to see, but in my experience large-scale open source applications (such as Linux, Open Office, Mozilla etc.) tend to "evolve" as many programmers contribute (often having different, even conflicting goals), rather than follow a designed path. This quite often leads to problems with consistency across the application, large size, etc.

With AROS and OpenBEOS this isn't so much of a problem as they have a blueprint to follow. Sure there are examples of good and bad in both proprietory and open source applications (Windows is a badly implemented proprietory system, for example), but I don't like the way large open-source applications tend to evolve.

This is, of course, just my opinion.

I should note that i'm not against Open Source per se, but I don't think as a development model it is the ideal one for an AmigaOS-like system.

And again, I do recognise that AROS seems to be pretty cool, and I do not underestimate the value of the code it has contributed to both MOS and AmigaOS. AROS has benefitted from having a blueprint to follow, and from being in such a small niche that it has remained withing the control of a select group that can shape it.

[xeron]

If we obtain the AmigaOS in the bankruptcy this is what we will do:

I can't believe you posted that sentance in quite that wording, to be honest.

3. Give the AmigaOS to AROS!

You can't "give AmigaOS to AROS". Releasing the source code to the official AmigaOS would infringe all sorts of patents and licenses, or so i'm lead to believe.

[xeron]

Well, its nice to some.

Personally, while I have nothing against AROS, I don't really think it should be the official successor to AmigaOS; having read the AmigaOS4 featurelist, *if* OS4 actually follows it, I think I would much prefer that.

Maybe if AROS copied some of those features, i'd agree  :-D

[xeron]

downix wrote:
Those patents and licenses cover specific areas of AmigaOS, such
things as the font engine and ARexx.  Also the right mouse click
patent.

I can't comment on the font engine, but AFAIK, the exact same ARexx binary from 3.x will be included in OS4, and I presume they know what licensing needs to be attained to include this.

As for the right-click patent, I was under the impression this was owned by Gateway, licensed to Amiga, Inc., and sublicensed to Hyperion.

Actually, for the font engine, I presume you are talking about the bullet engine, not the whole Amiga font system, since the bitmap support is very much created to be easily done almost entirely by the Amiga's blitter, and hence would have been done inhouse.

[xeron]

Who needs that old thing, We have much better systems now.

I know, I was just not sure what was meant by "font engine", is all.

[xeron]

Indeed, anyway what about these so called "Feature" that AOS4 has that AROS doesn't have.

First, my understanding of AROS is that its goal is to first provide the functionality of OS3.1 + MUI and only THEN to improve upon it. AOS4 already *is* OS3.1, and they are adding improvements. Has this changed? Has AROS reached its primary objective?

OK, here is a list of features which are currently touted for inclusion in OS4 that I think make it more apt for a continuation of the "official" AmigaOS.

Note, this doesn't stop AROS being a fine product in its own right.

Please correct me if AROS already has any of these, or has them marked to be included any time this year. (i think "this year" is a fair limit, since the majority of people expect OS4 to be delivered this year).

1) A JIT 68k emulation for starters. AROS can only run 68k apps on 68k.
2) Unified input system. AmigaInput looks really good.
3) Resource tracking for memory, message ports, messages, and semaphores
4) Unified MMU interface (difficult to implement for a system with multiple CPU targets.)
5) Limited memory protection. Protection for critical memory areas; Kernel memory areas, all code areas, unused memory.
6) WarpUp emulation
7) Symmetric Multi-Processing (SMP): support for multiple CPU's
Multithreading
9) New library model with dynamic interfaces depending on the methods used to invoke the library by the application
10) Fully virtualised address space
11) 3D drivers (Warp3D followed by Warp3D Nova)
12) The application library (something that should have been there since at least 2.x)
13) Roadshow (although I understand other Amiga-like OS's will get this later)
14) Mount Rainier support
15) AmiDock. I was never a big fan of Docks, but AmiDock in 3.9 won me over. 4.0 looks great, and from what I hear will be quite well integrated with the OS (for example application management through application.library)

Extra disclaimer  :-D
I do like AROS, I appreciate the value of the code that has been contributed from it to MOS and AmigaOS, and I also want to see AROS do well.

Of course, if OS4 doesn't live up to its featurelist, this is all irrelevent anyway  :-D

[xeron]

ARexx was third party, that can be relicenced. AROS chose to remake ARexx from scratch, Thus our ARexx is called Regina. And yes it is fully compatible

Of course it is fully compatible; currently AROS 68k is not being worked on, so all AROS applications require a recompile, and thus will work happily with any ARexx implementation.

The problem is, some AmigaOS programs have old amiga.lib routines actually linked into the binary, which would break. You would notice some problems with AROS 68k, and software that hasn't been recompiled.

This means that Hyperion, or anyone else for that matter, making an AmigaOS binary-compatible operating system would lose some compatibility if they scrapped the current ARexx implementation.

[xeron]

1) A JIT 68k emulation for starters. AROS can only run 68k apps on 68k.

Ok this is one we don't have any real plans for, but we have made sure it can be added is so desired, and the hardware is capable.

So AROS won't have it this year then? (which was my point). This is the big one, the main fundamental reason why AROS isn't the best contender for the continuation of the *official AmigaOS*; you can't run Amiga apps on it without recompilation, except on 68k. Nobody thinks 68k is the future. :-D

Can you imagine the confusion if AROS became the official AmigaOS? New users getting a new AmigaOS machine, downloading software from the internet "for the Amiga", only to find they can't run it because AmigaOS doesn't run 68k apps? And this would be 99.999999999% of the software out there!

2) Unified input system. AmigaInput looks really good.

I think our Prefs system is certainly over a very high standard.

"Unified input system" has nothing to do with the preferences system. I suggest you read the CAM article that has since been made public. (its on os.amiga.com).

3) Resource tracking for memory, message ports, messages, and semaphores
Yup, AROS has RT.

Nice

4) Unified MMU interface (difficult to implement for a system with multiple CPU targets.)
Difficult one, as we have had many disscusions as to how this is best done, and since there are so many different types of MMU, an mmu.library has been considered.

But it won't have one this year?

5) Limited memory protection. Protection for critical memory areas; Kernel memory areas, all code areas, unused memory.

MP has been talked to death,,, we can add limited MP (as you mention) with realative ease, but old APPs won't like it or care about it, which totally defetes the point. And limited MP is only one step away from useless.

Memory trashing is a common cause of crashes on the Amiga. Limited MP won't break existing apps (since it doesn't limit memory on a per-task basis for legacy apps), but it will help stability. Admittedly, OS4's intelligent stack management will have a much bigger impact on that stability, but even so. Laying the foundations and all that.

6) WarpUp emulation
Not applicable. This has never been a goal, no doubt, on a PPC it could be easy, very easy to write a warpos.library.

Indeed. Its still a plus point for OS4, and something I want to see.

7) Symmetric Multi-Processing (SMP): support for multiple CPU's

Yup, much has be talked about. And is certainly planned.

Will it have it this year?

Multithreading

Hello, this is an AmigaOS clone :-D Of course you can multithread :-)

AmigaOS has never had proper multithreading (remember: multithreading is not multitasking). Therefore your quote is meaningless.

10) Fully virtualised address space

No, I don't think that is a planned feature, though I remember the PCI guys talking about something to do with this...

Its a very good idea; and very doable now that AmigaOS is migrating to another architecture and keeping JIT for legacy applications.

11) 3D drivers (Warp3D followed by Warp3D Nova)

Of course 3D drivers are being planned, though AROS has chosen OpenGl rather than Warp3D

OK. If they have them this year, I'll concede that point.

12) The application library (something that should have been there since at least 2.x)

I'm not sure what you mean... I'm sure AROS has an equivilent.

I'd ask the developers about that

13) Roadshow (although I understand other Amiga-like OS's will get this later)

Roadshow? if it can be compiled for AmigaOS 3.1, it can be compiled for AROS.

Yep. I'm sure AROS will get Roadshow. In fact, I'll concede this isn't one of the reasons OS4 is a better choice to continue the official AmigaOS.

14) Mount Rainier support

I have no Idea what that is, see comment 13

OK, well, its a good thing to have.

This is correct to the best of my knowledge   :-D

Great. I'm not contesting that AROS is, or will be brilliant. I'm confident that if it continues as it has so far, it will be fantastic. I'm looking forward to running it myself.

My point is that AmigaOS4, IF it comes out this year and IF it lives up to its featurelist is a better choice for the official AmigaOS for the reasons above, simply because it should, in theory, get those features out sooner rather than later.

So, really, I think AROS should stay called AROS, AmigaOS4 should stay as the next AmigaOS.

Anyway, this is all hypothetical until either AROS matures enough to be used as a main operating system, or OS4 arrives.

Doubtless many will call me hopelessly optimistic, but I like to think the latter will happen first  :-D

[xeron]

Kronos wrote:
To put it simple : There is NOTHING breathtaking
in the OS4-featurelist (or in the MOS one), these
list just tell what a OS needs as absloute minimum
these days.

So? The question was "what features will OS4 have that make it a better choice for the official AmigaOS", *NOT* what features does OS4 have compared to mainstream operating systems.

Besides, protecting the kernel, unallocated RAM, and certain other areas won't break legacy applications, and neither will giving OS4 native apps true memory protection (since the 68k apps can overwrite eachother all they like but not touch native applications). This limited MP isn't spectacularily useful from a legacy point of view, but is clearly a way to introduce proper memory protection for native applications.

The thing that would kill legacy apps is protecting memory on a per-task basis for 68k applications.

[xeron]

OK. I'm going to end this discussion here, since i'm tired of it, and i'm certain that i have put forward a strong case for my original statement about AROS's aptitude for being called the "next official AmigaOS" (in a purely hypothetical world) anyway.

I do not agree with the idea that the sandbox and "jam tomorrow" q-box approach is the best one, or the only way to introduce memory protection, but the relative merits of MOS vs. AmigaOS are so off topic its off scale. It is also a very tired argument.

Lets just agree that all solutions have their merits, the future looks bright, and leave it at that :-D