Official update to OS 3.1 from Hyperion

[Samurai_Crow]

@wXR
+1!

[Threader]

@wXR

+2

And Thread-Thrasher will be my next nick... (And that was strangely how i came to the nick in the first place, threads had a tendency to die when i posted... )
Cheers

[matthey]

Originally Posted by Kalamatee

Which CPU choice is not the problem, the ability to run on which ever is the current "flavour of the week" with little effort is - something AROS wins hands down over all the other "AmigaOS" flavours.

Supporting many CPU architectures has advantages like uncovering bugs and requiring more robust code but it also has disadvantages. The advantage of focusing on a narrow set of similar architectures include faster coding, better optimized code for the architectures, pre-compiled architecture specific executables are more practical making distribution of software easier and less cross-platform maintenance and testing/debugging as you described perfectly.

Originally Posted by Kalamatee

Understandably, the limited number of devs are also held up more with getting the code/ecosystem working across different platforms in the first place which is a considerable task/feat considering the amount of hardware/build environments it works on and with.

Is is easy to see the advantages of cross-platform development but I wonder if you see the disadvantages even as you speak of them.

Originally Posted by Kalamatee

That's not entirely true - or hasn't been as bad for the past few major versions. The main issues is the lack of effort to provide proper "amigaos" shared versions of the used modules, resulting in bloated statically linked binaries.

The problems with the header files are not as bad for GCC as it sets the standard. The default GCC standard is a sloppy C11 plus POSIX plus GCCisms that other compilers must follow for compatibility (most programmers have been programmed to use the default GCC slop mode rather than a C standard). Some other compilers don't want to put everything together into one big bloated compiler mess but it is difficult to separate out the pieces and add them back in in a modular and compatible way.

A statically linked POSIX library is more flexible and easier to use and develop with a cross compiler for multiple targets. Light POSIX using software would be better off with the static library while heavy POSIX using software would benefit from a shared library. I would want to get Frank Wille's logic and opinion on a shared vs static library though. He probably considered a shared library but ixemul.library is shared and he very much dislikes it but likes BSD.

Originally Posted by Kalamatee

AROS has had TLSF for some time now, and has used it as the default memory allocator on most platforms for the whole time, so it has been pretty well tested.

But probably not on the 68k because it is compiled for the 68000 and TLSF needs the 68020 BFFFO instruction to be efficient?

Originally Posted by Toni Wilen

When I can compile aros with "fixed" gcc (without too much pain!), I'll check if code looks better

How about compiling AROS for the 68020 and with TLSF and checking if the 68k AROS TLSF memory functions are using BFFFO as well? Also do some benchmarks (code size differences would be interesting too)? Anyone have a Vampire accelerator to test the performance gains with several times faster BFFFO instructions?

Originally Posted by Toni Wilen

Use of built-in register parameter support should also improve generated code quality (vs current really ugly inline assembly hacks..), assuming it enables optimizer to work more optimally.

Originally Posted by Kalamatee

My experience (and understanding from discussions on the mailing lists for GCC etc) is that it in fact has the opposite effect in a lot of cases due to forcing the compiler to work with specific registers/not letting it choose/reorder them, especially once the code is optimized.

Toni may be referring to (needs clarification) the way the compiler can more efficiently use the __builtin macros/functions compared to inline assembler?

https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html

He would be correct for GCC and they are much preferable to inline assembler for multiple architectures. I haven't looked at the AROS code but I would expect the AROS TLSF functions use these __builtin functions?

Kalamatee, you were referring to REG variables? In my experience GCC, vbcc and SAS/C all do a good job of register allocation without using them. I have not noticed much of a difference where used but they are probably just a hint like inline which GCC usually happily ignores. Compilers can lose some efficiency around inline assembler though.

Originally Posted by wXR

I am amazed and really proud of you guys. So what's next? Thrasher has a good idea there, let's start a thread about AROS improvements that we can help along with a bit of bounty money, time, and elbow grease. I'd love to see this momentum leveraged to the fullest extent.

Competition is a better way to put pressure on the Amiga road blockers than suing them which inflicts pain on you as well. The current precious holders are going to wake up one day and realize their precious isn't so precious anymore. AROS is already more advanced in several ways than AmigaOS 4. The Apollo Core is now booting AROS as well. Jens/Hyperion better have something good to announce which will unite the majority of the Amiga community or they may find themselves as just a smaller and smaller inconsequential minority.

[wawa]

@matthey

Supporting many CPU architectures has advantages like uncovering bugs and requiring more robust code

you say it..

but it also has disadvantages

perhaps. but aros is multiplatform by design, take it or leave it. the question is if advantages outweight handicaps. i would add aros has a number of futher features to offer. not even system features to speak of, but opportunities, as a handy infrastructure to build the whole system (os3 sources need yet to arrive at this point apparently), bunch of experienced coders, possibility to easier implement solutions ported from the outside world, whatever they might be.. stuff like that.

amiga os or os4 being locked away from potential contributors offer nothing at all in comparison. not even didactic values, no further generation of amiga coders will arise as long they cannot access und understand the code and functionality the system consists of.

But probably not on the 68k because it is compiled for the 68000 and TLSF needs the 68020 BFFFO instruction to be efficient?

its been a measure in order to warrant compatibility, intead to be distracted testing another optional subsystem. i think it compiles but im not sure anymore.

How about compiling AROS for the 68020 and with TLSF and checking if the 68k AROS TLSF memory functions are using BFFFO as well?

why not? i have not been succesfull to compile full aros with the 68020 flag, but some modules might do, some tweaking may be necessary. i can help you with setting up build system if you want to try.

Jens/Hyperion better have something good to announce which will unite the majority of the Amiga community or they may find themselves as just a smaller and smaller inconsequential minority.

i find it a bit annoying i must admit, that you continue to see aros only as measure to put pressure on the vendors you mention. i dont think its going to work out. aros should be seen as an opportunity in itself. anything else is a waste of time.

[alpine9000]

@wawa aros is great, but with the exception of the kickstart replacement it seems (at least to me) to be focused on the "next gen" section of the Amiga community (which is fine). But there is also a large "classic" amiga base and they would also like to see improvements in the operating system space.

So by focusing 100% on aros, that group would be left behind.

So unless aros would include support for classic Amiga (which seems unlikely based on the comments in this thread), the classic Amiga scene is left hoping aros competition provides a prompt for the current rights holders to finally improve the classic OS.

[wawa]

Originally Posted by alpine9000

So unless aros would include support for classic Amiga (which seems unlikely based on the comments in this thread), the classic Amiga scene is left hoping aros competition provides a prompt for the current rights holders to finally improve the classic OS.

what is the difference between "kickstart replacement" and aros (68k) as a whole? you seem to think that jason and toni only exclusively worked on kickstart part, as requested by the bounty. it is wrong. they werent that lame. they have commited towards a whole basically working system on amiga platform.

aros include support for what you call "classic" amiga (spit in the corner) which in fact is simply Amiga. check aros sources online for yourself, if you dont want to check them out. navigate to arch/m68k-amiga and here you are:
https://trac.aros.org/trac/browser/A...rch/m68k-amiga
this is amiga specific code. the rest, out of arch and except of ifdefs is cross platform. does it mean that it is leaving amiga behind? no. it means it is dragging it along. damn..
last week we identified a part of cross platform smp support code that caused some (i repeat: some) amiga applications fail, therefore it wasnt apparent right away, maybe because none is helping me with testing amiga-m68k target. kalamatee immediately reverted that commit. does that sound like leaving amiga behind???

aros is being focused on whomever contributes. if amiga (68k) people contribute or atleast test, it deserves its share. you dont, you get left behind. it is your choice, not evertyhing needs to be presented to you on a silver plate.

excuse me, my irritation, but i have to continously straighten misconceptions for people who dont even care to check if what they suspect is true. thats the reason we are currently left with threads of bragging about utopic deeds like open sourcing what is intentionally locked away, instead of real coding action, to improve given situation.

[alpine9000]

@wawa - sorry for cross quoting this from the other thread

Originally Posted by wawa

sigh.. it isnt meant for vanilla amiga 500 or 1200.. what features do you need with your vanilla amiga to have enabled beyond what is there and what provides best compatibility for existing software?

well, i imagine, you might want, cross kickstart compatibility for applications, say between ks1.x and ks3.x. aros provides this at lest to some extent, which is an additional feature in comparison to original. but you might imagine that it comes at a cost. memory requirement in this case. you need 1.5mb ram to be able to boot without the s-s, 2 mb to use anything effectively.

aros is currently compiled for plain 68000 but it wont fully boot there according to my tests, suspectedly due to some 68020 asm inlines. it wouldnt make sense at this point anyway, but it actually even might be possible one day.

I guess maybe I misunderstood you, but I initially read this post as it being unlikely that aros would be heading in a direction that would make it a viable OS replacement for a vanilla A500.

[wawa]

Originally Posted by alpine9000

viable OS replacement for a vanilla A500.

omg..



.....

whats wrong with the genuine os on vanilla a500?????????????????????....
what do you need????

[alpine9000]

Originally Posted by wawa

omg..



.....

whats wrong with the genuine os on vanilla a500?????????????????????....
what do you need????

The source code

[wawa]

if you need only an update to a simple device, like support for partitions and disks beyond 4gb fence, simply compile an aros module like scsi/ata device check if it is romable, get rid of (if any) potential posix dependencies and if it fits, compile it in your custom amiga kickstart using remus.

[wawa]

Originally Posted by alpine9000

The source code

have i not explicitely pointed you to the source code already ?

if you want to check it out via svn, there is full instructions to do so on aros portal. namely here:
http://aros.sourceforge.net/document.../compiling.php
scroll down the page:

> svn checkout https://svn.aros.org/svn/aros/trunk/AROS
> cd AROS
> svn checkout https://svn.aros.org/svn/aros/trunk/contrib

given you have all necessary packages installed you can then make a build dir and from there invoke configure, with some - your preffered - options, like:
--target=amiga-m68k
--with-serial-debug (if you want serial debug)
--enable-debug (if you want debug symbols compiled in)
--with-portssources=.. (if you want contributed archives and toolchain sources stored locally instead downloading them anew)
--with-gcc-version=6.3.0 --with-binutils-version=2.25 (if you want the neweat toolchain)

and then make (-s)

i can tell you how to compile a part of the source (a module) but if you want to use another (external) compiler, you will have to experiment on yourself.

[alpine9000]

Originally Posted by wawa

if you need only an update to a simple device, like support for partitions and disks beyond 4gb fence, simply compile an aros module like scsi/ata device check if it is romable, get rid of (if any) potential posix dependencies and if it fits, compile it in your custom amiga kickstart using remus.

It's not about any particular feature, it's about freedom.

You say that aros is not a good choice for a massive slice of Amiga hardware out there, yet you seem to be against testing the waters to see if it's possible to free the OS for all of those basic old Amiga's?

I don't see why this is contentious. I can't see that a GPL OS3.1 would change anything in the aros world? But it would sure change things in the "vanilla A500" world.

[wXR]

@alpine

The one thing it would absolutely change in the AROS world would be the "R" part. i.e. no more need for the "clean room" approach. Still a goal I am pursuing.

[ross]

Originally Posted by wawa

--with-portssources=.. (if you want contributed archives and toolchain sources stored locally instead downloading them anew)
...
--with-gcc-version=6.3.0 --with-binutils-version=2.25 (if you want the neweat toolchain)
...
i can tell you how to compile a part of the source (a module)

Hi wawa, we are really off-topic, You can please open a new thread about all this AROS compile options?
(is not 6.3.0 WIP and have issue with pthread?)

Thanks,
ross

PS: yes.. I've not RTFM, there is information, but scattered everywhere...

[wawa]

Originally Posted by alpine9000

It's not about any particular feature, it's about freedom.

you welcome, and now?

You say that aros is not a good choice for a massive slice of Amiga hardware out there, yet you seem to be against testing the waters to see if it's possible to free the OS for all of those basic old Amiga's?

im not against anything if it was realistic. what have you been able to do, in order to achieve it?

and i ask again, what do you practically expect of a better system, to demand it for a vanilla a500? this all costs effort. im talking about practical goals not some fancy, you brag on forums about.

I don't see why this is contentious. I can't see that a GPL OS3.1 would change anything in the aros world? But it would sure change things in the "vanilla A500" world.

yes. again, what would that change precisely? you have your pirated sources foating around for months, maybe years.. have it changed anything?

[wawa]

Originally Posted by ross

Hi wawa, we are really off-topic, You can please open a new thread about all this AROS compile options?

open one yourself intead constantly beating me to answer here.

(is not 6.3.0 WIP and have issue with pthread?)

it is wip and there certainly are issues, known and unknown but it compiles and runs the whole aros 68k boot iso and most of the contribs, perhaps even ports.

[wawa]

Originally Posted by wXR

@alpine

The one thing it would absolutely change in the AROS world would be the "R" part. i.e. no more need for the "clean room" approach. Still a goal I am pursuing.

you can compile aros today, without that "r" if it suits you. its your own responsibility if you want to throw the fortune on a single character, even if it belongs capitalized..

[wawa]

sorry guys i need rest. all this arguing is sensless. use and contribute to what you want. it is much more rewarding that arguing on forums, as long as you are able to. all day lost because vampire team got aros booting and scisors got awake. good i got real job too.

[alpine9000]

Originally Posted by wawa

im not against anything if it was realistic. what have you been able to do, in order to achieve it?

and i ask again, what do you practically expect of a better system, to demand it for a vanilla a500? this all costs effrot. im talking about practical goals not some fancy, you brag on forums about.

yes. what would that change precisely? you have your pirated sources foating around for months, maybe years.. have it changed anything?

I don't really see this as a big deal.

A couple of people are looking into if it is possible to get the current rights holders to work towards an open source release of OS 3.1.

It will probably will go nowhere, as my guess is no-one is legally in a position to actually release the code (or if they are, they don't want to), but I don't think we should sidetrack people from checking into it just because aros exits as an alternative ?

AROS looks to have a great future. Vampire/AROS will be a great platform. But some of us love our basic old Amigas, and a big part of that is the original operating system(s). An open source release of that operating system would be an amazing thing for people like me.

[wawa]

Originally Posted by alpine9000

But some of us love our basic old Amigas, and a big part of that is the original operating system(s). An open source release of that operating system would be an amazing thing for people like me.

if you like your system as is, then im not sure what you want to gain or change about it. let alone i dont see why you feel threatened with aros opportunity. as if it would change anything for you if others would choose aros themselves.