Roadmaps and R1

[_AP]

@PulkoMandy

Maybe just throw away the whole kernel and run our userspace on Linux.

Ew. No.

---

[bitigchi]

Even from 2019 I can tell. Haiku is definitely rock-solid once you get the right hardware. However, there are three big blockers from my point of view.

• Data safety; we need a rock solid file system that won’t trash people’s data and their brand new SSD with erroneous writes. Plus, some sort tight-coupled disc encryption that will keep people’s data safe. Add some sort of sandboxing into the mix.

• Hardware acceleration, it is no good when people cannot simply make use of their shiny hardware. I would also love a composited interface that gives a smooth and solid UI feel.

• Efficient and rock-solid web browsing. Preferably WebKit with tight system integration, or improved Qt compatibility with Falkon as stopgap. But I think as above bullet point is fixed, the WebKit integration will greatly improve to a point which will make using Web+ a pleasure. The application itself needs a lot of work though.

This is my humble two cents about what a ready Haiku would look like. Anything less than this does would still be beta for me.

[suhr]

For me, there are other blockers:

• No bluetooth audio. I can’t use my wireless headphones on Haiku
• Issues related to input server. KeymapSwitcher is a hack, and changing keyboard layouts with CapsLock does not work properly. Touchpad scrolls in a wrong direction, as well as horizontal scroll wheel on my mouse. No support for something like Compose key to enter symbols. And so on, it’s very painful to use an OS that does not support input properly.
• Power management. Haiku makes my laptop hot.

For me, 3D acceleration is much less important than anything above.

[_AP]

Is someone having their data trashed by BFS? I’m actually curious. I have several BFS drives (nothing important not backed up), and haven’t had a problem. That’s not to say someone isn’t having a problem. But encryption sucks. That’s where I’d be most concerned with losing data.

I would argue that the biggest block would be hardware support. Trackpads, network adapters, audio. The operating system itself is so far for me been very useful where hardware allows. WebPositive hasn’t been the best for me, and I only recently discovered Falkon. Falkon seems more stable.

Im not expecting a lot of graphically intense work on Haiku, not that those aren’t bad things, but for most people using Haiku, it’s a non issue.

[SantiagoLema]

Just noting here that hardware acceleration isn’t only about 3D apps or games. Apps such as Krita aren’t really useable when using paint strokes. In VESA mode.

[Zenja]

Regardind 3D acceleation, there has been work done by X512, who actually got HW acceleration working for Radeon Sea Islands hardware (pre Vega). See this thread:

VideoStreams: Media Kit-like 3D hardware acceleration kit for Haiku

Copied from this post: I made an illustration of my 3D hardware acceleration architecture proposal: [Hardware graphics acceleration architecture] It introduce new native Haiku API for rendered buffer producers and consumers. It works in a way similar to Media Kit. Application that will render 3D graphics with OpenGL or Vulkan will expose BufferProducer interface that can be connected to BufferConsumer interface providers such as screen of compositor that will mix multiple inputs to one output…

Vulkan lavapipe software rendering is working on Haiku

radeontop: [screenshot240] [screenshot241]

[PulkoMandy]

Ok, this thread is turning into a feature wishlist all over again.

The R1 goal is very clear. A replacement for BeOS R5. There should be no new features. The goal is not “a fully usable OS in today’s environment”. Where did you get that idea from? If we set that as a goal, we will never reach it, because “today’s environment” changes faster than we can handle.

So, it’s all compromises and arbitrary decisions. Not for the R1 goal, but for what people actually work on. Typically, it will be fixing bugs we stumble on, adding features that we need ourselves. These things may or may not be part of the R1 goal. It doesn’t matter.

For me the OS is already usable. I’m running it on my main machine and I do most of my stuff with it. I’m not alone in that case. So, if the goal was “an OS that’s fully usable in today’s environment (for some people on some hardware)”, the R1 release would already been shipped.

[trungnt2910]

@PulkoMandy

Maybe just throw away the whole kernel and run our userspace on Linux.

Funnily, there has been talk about using HyClone for testing 3D acceleration applications…

[Begasus]

@bitigchi

Data safety; we need a rock solid file system that won’t trash people’s data

Happens to the best (read KDE) recently.

[bitigchi]

That was not file system related, it was an arbitrary code execution error. Still something we need to mind as well.

[nmatavka]

@PulkoMandy

All options are still on the table. Throw away all our drivers and import Linux ones with a compatibility layer. Keep our native modesetting drivers and port only the userspace parts from Linux. Maybe Mesa, maybe Vulkan. Maybe just throw away the whole kernel and run our userspace on Linux.

HELL to the no. If we’re going to be discussing alternative kernels, then maybe the Zircon kernel from Google Fuchsia. That’s the only one I would support, because everything else is a kludge (Linux from the 80’s, BSD from the 70’s, etc.)

[_AP]

@nmatavka

then maybe the Zircon kernel from Google Fuchsia. That’s the only one I would support, because everything else is a kludge (Linux from the 80’s, BSD from the 70’s, etc.)

Ew. No.

---

[trungnt2910]

@nmatavka

If we’re going to be discussing alternative kernels

Discard all the drivers, compile the Haiku kernel as a Windows NT driver, then forward all hardware-related stuff to Windows NT.

@_AP

Ew. No.

At least that solution still uses the Haiku core.

[nephele]

Finally merging ReactOS and Haiku . : )

[dcatt]

@PulkoMandy

Maybe just throw away the whole kernel and run our userspace on Linux.

Though I love Linux as much as the next guy or gal, please please no! We don’t need yet another Linux distro.

[nmatavka]

@_AP

Ew. No.

That’s very informative(!) Considering that Travis Geiselbrecht, the author of the NewOS kernel, which was forked to create the Haiku kernel, also wrote the Zircon kernel, can you share whatever wisdom inspires such feelings of revulsion in you(?)

Or is this just the textual equivalent of a fart?

[PulkoMandy]

I thought the inclusion of Glide drivers for Voodoo 3D cards from the early 1990s was enough to hint that my list of suggestions was not entirely serious. Not sure why the one with using the Linux kernel is attracting so much attention then. Where are the comments by people offended by the fact that I plan to waste my time on legacy hardware that no one owns anymore?

[RogFanther]

Parallel-port connected Zip Drives ?

More seriously :
A good enough roadmap wouldn´t be to first reach Beta 5, then see what can be done next ?

[cb88]

I’m just going to point out that the best developments in Haiku in recent years have been due being less isolationist… sure there are warts but Haiku actually can run a large library of software now, due to implementing compatibility layers.

The same goes for drivers as well as applications, Haiku uses FreeBSD’s network drivers and it won’t have much choice but also use Linux’s GPU acceleration drivers because that is where all the open source work is happening, the up side to this is that nearly all of the code in that area of the Linux kernel is dual licensed MIT so compatible with Haiku license wise.

Then there are also compute specific drivers showing up on the scene for things like XDNA…GitHub - amd/xdna-driver which is an driver that lets you run the Xilinx XRT on recent AMD CPUs.

[dcatt]

Why do you think Google’s kernel is the answer?