AFAIK all the things needed for a modern kernel is already there in Quark. Sure, you get a lot more drivers in Linux, but will it still be it's own thing? I don't think so. I think it's the same reason why Haiku voted against going for Linux too. It would be less authentic and less original.

As I understand it currently the Quark microkernel is used to boot the hardware and host the A/Box and not much more which isn't too dissimilar to how Amithlon used the Linux kernel.

Assuming MorphOS NG will be more or less the mythical Q/Box that makes use of the advanced features Quark provides rather than just playing host to another OS like it does now, then there is always the ossibility of adapting Quark to be able to use Linux drivers in a similar fashion to the Genode OS.

http://genode.org/documentation/release-notes/14.11#Intel_wireless_stack

[ Edited by Intuition 08.05.2016 - 17:13 ]

> going Linux as the kernel doesn't mean you cannot write an OS that
> doesn't feel like an Amiga

You got entangled in your negations here :-)

Linux, TempleOS, Haiku, and GenodeOS, damn.

How about if MorphOS (NGNG) is based on...MorphOS?

Jim. for sure MorphOS should (and will) be based on MorphOS. But looking to alternative solutions may help to make thr right decisions. Evaluation of these solutions will give a vote on what to do and what not to do.
As far as I know the history and the developers I ampretty convinced they will not go for a Linux base. But if they eventually did, it would not be the end of the world or MorphOS would instantly beome YALD. It could stay pretty authentic.

That said I vote for not using a *nix kernel or an other alien kernerl, but let Laire et al do the real hard job themselves

Quote:

You'd not use MorphOS if they added the ability to use Linux drivers to the kernel?

Zylesea, this is one area where I feel we have a real advantage.
So my (cough, cough) 'mild' form of fanaticism is on display again.

Top to bottom, I am really fond of this OS.
It incorporates some of the best of our legacy OS with ideas that were very cutting edge at the time of its development.

AND a Linux kernel is philosophically the opposite of our current structure.

It scares me a little when people suggest we scrap concepts that are a)relatively unique and b)work really well.

Quote:

Jim, What I think Intuition and I are trying to say is NOT that we want to switch the kernel to Linux, but if possible, adapt the MorphOS kernel you are so fond of, so it can use Linux drivers. Maybe this is not possible, but until someone more knowledgeable than myself steps into this conversation and says what we are suggesting is about as possible as getting the AmigaOS4.x team and MorphOS team to join forces, I will continue to support the idea of incorporating Linux drivers into x64 MorphOS, to reduce the workload our developers have to do now and in the future.

It may very well be "pie-in-the-sky" dreaming, by the uneducated concerning OS design (myself), to suggest or wish for a new MorphOS that can somehow also use Linux drivers, and in a perfect world, magically also contain a wrapper, or some other coding to allow us to run Linux software.

I, like you, also do not wish to have MorphOS turned into anything resembling Linux, but if there is a way to accomplish the above without destroying what MorphOS has been based on from the beginning, it would be great to instantly have the entire Linux software library available, in addition to tools that make writing new MorphOS native software easier.

Like I said, this is probably impossible, but until I see it written, or hear it spoken from someone I know is likely to have the knowledge to answer this question, I'll remain ignorantly hopeful.

If I am going to dream, might as well dream big!

They really are radically different types of structure David.
Many elements incorporated into the Linux kernel don't need to be there.

Placing essentially all drivers above the kernel (as a microkernel OS would) just makes sense to me.

> Linux, TempleOS, Haiku, and GenodeOS, damn. How about if MorphOS (NGNG)
> is based on...MorphOS?

Nobody suggested future MorphOS should be based on Haiku or GenodeOS. And the TempleOS suggestion was an obvious joke.

@Amigadave

And what drivers would that be ?

The interesting parts of the GFX-drivers are in binary blobs and we allready have a quite good base here,

Basic stuff like IDE/SATA and network doesn't really change that much over time, so it's "just" a matter of adapting existing drivers (and choosing HW where those components are related to ones allready supported).

Only leaves the real basic stuff, scanning the device tree, setting up the northbridge and so on.
Don't think twisting a linux-kernel here will lead to optimal results.

> The interesting parts of the GFX-drivers are in binary blobs

Couldn't these be used on MorphOS/x86(-64) with Linux-compatible kernel as well?

Besides:
http://en.wikipedia.org/wiki/Free_and_open-source_graphics_device_driver#Free_and_open-source_drivers

> and we allready have a quite good base here

Nearly as good as the Linux base?

Quote:

Well currently all drivers (Someone correct me if I'm wrong) all drivers in MorphOS run in the same space as the Exec kernel hence the desire to move to using Quark instead.

Genode uses a microkernel (L4 or a few other options/variants) and everything else including the Intel wireless drivers from Linux run outside of the kernel.

I don't see why it's an issue for you Jim.

Quote:

I will have to look at that, but they must have really struggled to pull that off.
And it would not be Linux based.
Linux IS the monolithic kernel that Linux based OS' run on top of.
Create an OS with a different kernel and its not Linux.

Further, do some of you (Andreas?) think we have a developer base that can create an OS that can utilize Linux drivers without simply defaulting to Linux?

You are asking for an enormous amount of work.

Quote:

There's not that many working on Genode and they are mostly CS students and their professors IIRC. Our developers are more than capable I reckon but whether they desire to add this capability is another story. It would be great (for us at least!) if they do though.

Even adding support for BSD drivers like Haiku and others have done would give us a lot of extra hardware support for "free" and wouldn't be a PITA licence-wise compared to using Linux drivers.


[ Edited by Intuition 09.05.2016 - 12:38 ]

> they are mostly CS students and their professors IIRC.

Genode Labs is a commercial company, founded in 2008 as a German GbR and transformed in 2010 into a German GmbH. The founders stem from academic research background.

http://www.genode-labs.com/company/index?lang=en

Double post

[ Edited by Raf_MegaByte 09.05.2016 - 15:51 ]

Quote:

But Apple uses XNU Darwin Kernel not a generic Linux kernel, and also they modifies the Kernel for their purposes... It is not a plain *nix kernel anymore.

BSD parts embedded into OSX provide other functions than Kernel, for example the roundrobin task scheduler in OSX is based upon that original BSD task scheduler.

Apple also made a great job of adapting the hybrid system born from the mix of BSD + Darwin Kernel to make it appeal as legacy Apple System OS, i.e. the new operating system uses HFS+ filesystem instead of EXT3/4 from linux world, it includes Apple made GUI and windows manager look&feel aspect equal to match legacy System 6/7/8/9 OS versions, in order to make users feel the same experience of their ancient operating systems, and in the end they adapted the naming system of elder MacOS versions and APIs to fit the new hybrid XNU/BSD system they created.

At these conditions I will accept an Amiga Operating System with an Alien Kernel, that is adapt this alien kernel to use more practical and lightweight Amiga-lookalike Windows Manager, similar to AmigaOS or Ambient, using superior AmigaOS/MorphOS filesystems like ICEFS, and Amiga naming system and Directory Structure (and I intend no crappy linux naming like linux /dev/sda1, dev/sda2, etc, but solid DH0: DH1: etc.,) and human names for directories (like System, Utilities, Devs, that are plain explicative in comparison to Linux developers named for-geeks-only 3 letters like for example USR, Bin, Dev, or 4 letters "Home"...)


I repeat myself... Only at these conditions that must include a complete "Amigafication" of the hosting kernel I could accept the use of an alien kernel ported into Amiga...

But if someone plans to make an Amiga environment hosted upon a complete running Linux environment like as with Hosted AROS, with all Linux inherited difficulties that make it unsuitable for end-user and all Linux hydiosincracies, like enormous unusable GUIs (and do not forget the worst: Linux is suitable for geeks only), then if THESE are the plans it is better you bat your hands with a stick renouncing, and thinking twice what you've intended to do...

@Jim

That said, I want to add that I agree mostly with Jim when he said we already own Quark Microkernel still waiting for more adoption of its native featueres, deprecating as most as ancient Amiga legacy as possible, that is getting rid of 1,5 Gigabytes RAM limit, 128Megabytes for Videocards RAM limit, and so on...

[ Edited by Raf_MegaByte 09.05.2016 - 16:11 ]

Quote:

I beg to differ.

Quote:

Students with venture capital then. ;)

> Students with venture capital then. ;)

They stopped being students (one as PhD, the other dropped out as PhD student) in 2008 right before the formation of the company.

http://www.genode-labs.com/company/founders?lang=en

Quote:
I agree, and that's what I meant by using another kernel while keeping what's good in MorphOS/Amiga.

But most people seem to think that using a Unix kernel, means Unix/Linux filesystem, Unix bash + Unix Command lines, XWindow, Unix Windows System, and so on...

Except that you missing my point.
What IS good is what differentiates it
Otherwise, why not use something more mainstream?



[ Edited by Jim 15.05.2016 - 12:49 ]

Quote:

Ahem, I beg your pardon... are you responding to Leo or to me?

[ Edited by Raf_MegaByte 10.05.2016 - 03:16 ]

Quote:

I probably don't completely understand your point of view Jim, but I also think you might be dismissing my point without thinking of the advantages in the big view, or long run. I am not a developer and don't have any idea how much work is involved to modify the MorphOS Quark kernel to be able to use Linux drivers (lets just use video card drivers as an example), but perhaps spending the required programming time and effort now would be worth it, if it means future Linux drivers would be easily incorporated into the new x64 MorphOS.

If MorphOS is going to have a long life, then doing the extra work now will pay off by saving more work in the future, when new drivers are needed.

The way I see it is if it is possible to modify the MorphOS kernel so it can use Linux drivers (inside the kernel, or outside of the kernel so it can remain a "microkernel"), it will be like adding dozens of Linux programmers to the MorphOS Dev. Team, because those Linux programmers who work on Linux drivers would be working on something that could be included in future MorphOS versions.

[ Edited by amigadave 10.05.2016 - 00:20 ]