Opinions how to get to R1

It mentions 3 familiar issues:

  • “security isn’t one of its core strengths”
  • “it still lacks hardware 3D acceleration or video playback”
  • “the built-in web browser WebPositive is rather limited… (but there is) the GNOME browser, and (experimental) Iceweasel”

If these were addressed to eliminate concerns, you would have an R1 - and LOTS of media coverage - and new users trying it out ! … “it is awfully close to being ready to be a daily-driver OS”

3 Likes

For R1, further security enhancements are not planned, as those would require a lot of (incompatible) changes to the API. This will probably be an important aim for R2.
Patches are welcome for the other shortcomings :grin:

11 Likes

Patches are welcome for security as well :slight_smile:

(We can always keep bigger changes disabled and experimental as long as it doesn’t break things…)

2 Likes

Well, there’s definitely parts of the Be APIs that we can’t change without breaking API compatibility. But there are still more “security enhancements” we can do without breaking API compatibility, like adding missing permissions checks to syscalls (usually won’t affect apps running as root, so not an API break, technically), restricting who can call private APIs (e.g. installing debuggers), and so on.

5 Likes

Sure, yeah, I did not specify “further” that well :slight_smile:
Those changes are always welcome, but I hope that we already have most of those implemented – if not, I would count most of these as bugs, not security enhancements, even though they do just that.

1 Like

Please don’t take away the experimental networking API v2 with BUrl and BHttpRequest - these are really handy already :blush:

1 Like

Can’t do patches unfortunately, only suggestions:

  • Security: there are good config utilities already - Preferences , Process Controller, and Superprefs (in depot). With those - and any more to add - included in the default install, the release notes could say that Haiku has “security-by-configuration (no command line required)”. This would be acceptable as configuration is the basis of OS security anyway. And it would be attractive to new users who don’t like using the CLI.

  • Browers: you could offer users to download a “browser pack” of all the other browsers at once (if you don’t already). This could aleviate some of the browser anxiety.

  • 3D HW acceleration & video playback: if this is difficult to achieve, maybe it would be worth paying a developer to do it.

So, then, an R1-worthy release would mainly depend on the 3D acceleration work. And, so, it could be achievable within a year or two. Otherwise, if beta releases keep adhering to a high standard - that reviewers don’t see a need for - then an R1 will just keep getting pushed further and further away - and the potential to have new users & devs on board will be slowed.

1 Like

How would you be able to reasonably estimate the amount of time 3D acceleration would take? It would require many changes throughout Haiku to even work, based on prior discussions here between Haiku developers. Even with a paid developer to do it, I doubt 1-2 years would be enough to get GPU-acceleration to a workable state unless focused on a single GPU model maybe.

There was one stalled effort for AMDGPU and even then, it only supported a Southern Islands generation AMD GPU; it was a good showcase of how difficult such an endeavour is. There have been extensive discussions on how to implement GPU acceleration on Haiku, which AFAIK still haven’t been settled on yet. Same with implementing compositing on Haiku, too.

1 Like

I don’t think 3d acceleration should be a requirement for R1.
Maybe 2d video acceleration (something similar to xv). I think beos 5 had something like that.

But beyond that, I think it should be set as a goal for r2 or 1.5.

R1, I think, should focus on compatibility and stability. And have good enough performances to be usable as a daily driver.

In my opinion, we are almost there!

I tried it successfully on 4 computer.

I was able to browse the modern web on all the 64 bits computers I tried on (3 out of 4)

The main thing holding an R1 release for me, is the couple of crashes and oddities I have regularly.
And still some (boot preventing) bugs with some hardware.

I have access to a lot of old computers (close to a hundred - from the pentium 3 area to the Core i7 6xxx - towers, desktops and laptops).
I can test haiku on all of them, if you think it can help accelerate Haïku’s development in some way?

3 Likes

we do need a kind of haiku-box.

special CPU for haiku
special GPU for haiku

and bulabula…

:stuck_out_tongue_winking_eye:

3 Likes

2D HW acceleration doesn’t exist on modern GPUs anymore, as this is typically just emulated with 3D acceleration.

1 Like

I am still exremely confused why people think 3D acceleration is so important that it would be worth delaying R1 by a few years or so. I don’t understand, besides video games, what do you all need that requires 3D acceleration so much? Or is it just another thing on the endlesslist of things people keep coming up with to justify that Haiku is not ready?

We had LibreOffice at the top of that list for a few years. Then it was wifi. Then it was the web browser. Now it’s 3D acceleration (on its own, as a checkbox item, without any example of practical usage).

Personally I woulo much prefer to have sound on my laptop, and dual screen support, rather than 3D acceleration. Then I could listen to my music on Haiku while working, instead of needing a second computer. And after that I would probably want a good email client, and a eorking touchpad, and then maybe suspend to ram support so I don’t have to shutdown the laptop whenever I move it. But I also think none of that really needs to be in R1

25 Likes

you are so right

1 Like

I think also that sound is before 3D acceleration for sure.

It’s not an issue for me for the moment, because my current trick is to use the audio jack I have by chance on my old NUC.

But if I hadn’t any sound at all, it would be a real issue for my daily usage (no video, no music, no sound in retrogames… no future :stuck_out_tongue: ).

Email client : yes, it would be nice to have a real client! I would like to be able to read HTML emails and my current tests weren’t 100% successful with my current email providers.

Regarding 3D acceleration, it would be a “nice to have for the moment”, I’m not sure having video games on Haiku is a real priority.
You can already play great retro games with Haiku (RetroArch or DOSBox) based on SDL, and if you need to play the latest games, don’t use Haiku it’s not its main goal :slight_smile:

“Nice to have” for 3D is because it could be interesting for me to have more fluid rendering under Blender, but it’s only for my specific usage.

That’s why I agree with PulkoMandy.

By the way, Firefox (Iceweasel) was a great addition from last year, which proves that Haiku is not a “toy” OS :slight_smile:

2 Likes

I didn’t know that.
The last time I checked was on my GTX 1080, video display still used less CPU with XV than OpenGL on Linux.

That may has been just an implementation thing.

Maybe the XV path is just better optimised.

Well, that’s even better if 2d acceleration is not on the checklist at all.

I think that 3D hardware acceleration should be a deliverable for R1 stable. At least get it working for one graphics card and take the path of least resistance to get there whatever that may be.

Just for reference, could you elaborate on the things in a good email client ? Like, for example, things that Beam doesn´t do ( yet ) ?

Well, what is its goal?

No, I’m serious. The goal used to be to provide a drop-in replacement for BeOS R5, We blew past that a long time ago IMHO. Haiku runs on more hardware (and runs more software) than BeOS ever did, it has sensible and innovative package management … It’s stable: I haven’t seen a KDL in months.

Refresh my memory, anybody. If the goal was to provide a replacement for BeOS, what is still missing, functionally speaking? What did BeOS do that Haiku still does not? Sure, there are things we would like to add. I’ll be first in the queue with suggestions. But in terms of the original goal, what’s stopping us from declaring victory and putting R1 out into the world?

Or is the goal now to compete with Windows/MacOS/Linux? THAT is a whole different kettle of fish …

4 Likes

I have done that already, but I can do it again:

  • Support for threading (grouping messages in the same discussion together)
  • Don’t create thousands of files locally, I don’t need them, I want my email to stay on my mailserver (I have 10+ years of email in my mailbox, and I don’t need to download it all)
  • Reliable synchronization with the IMAP mailbox (so I can delete mails, maybe move them to different IMAP directories, etc)

Hardware accelerated video decoding is another thing, not related to 2D acceleration. People get confused about all the things that a GPU can do for you, and it’s a lot of separate things. It used to be that each of them had a specific “fixed pipeline” processing support in a graphics card, in modern ones it tends to be more generic and programmable.

Functionally, nothing. The roadmap to R1 is a few hundred bugfixes and no new features (you can easily check it on the bugtracker). We reached that point in 2018 and it allowed us to enter beta stage.

Some of them are interesting only if you use these parts of the system. For example, media encoding through the media kit is completely broken, and has been for years. Some are subtle and not too important. Some are hardware specific.

1 Like