Is it possible to focus on a small amount of fully functinonal machines compatible with Haiku?

I’d like to see some real focus on just 5 - 8 main compatible and completely functional hardware machine setups for Haiku/BeOS.

I like BeOS and wanted to grow up using it more, but we didnt get that for some reason.

I’m a bit upset, that there is no management over the discussions the developers had, regarding, this weird “trying to take over windows on every pc possible vibe/feeling” i get from using haiku.

Focus please on 5 - 8 completely functioning and compatible machines and grow it out to say 19 after 4 years.

If you dont stop trying to program compatibility layers/drivers for so many disfunctional machines, (made that way because of Gates lack of aptitude to stop being a bully.)
there may not be a longevity status for HaikuBeOS.

I guarantee, after 14 months from today, if you dont stop and think and approach what I’m saying as a young men / women / adults and do something about focusing on specific custom hardware (like apple does) Bill gates will take over the OS, and there will be no more Haiku.

First problem with this - the sales lifespan of a specific model of computer is usually about 6 months.

Second problem with this - Bill Gates is not involved in operating systems anymore.

2 Likes

Would you buy one? For the record, this has been passing by for years. :slight_smile:

Each of the developpers is busy working on their own machine for the most part. That is a small number of developers (especially if you count only the ones who know how to write drivers) and so a small number of machines already.

What you suggest, essentially, is that we should close and ignore all tickets about any other hardware. To me (and to many of our existing users, I guess), this is not acceptable.

What changes in 14 months? Because we have been doing this for the past 25 years, and this still hasn’t happened. And I think Bill Gates is retired from Microsoft for most of that time, and busy with other things.

If you want, I can tell you which laptop I use: it is a Fujitsu Lifebook U7311. I spent a lot of time looking for a model that fits my need. If anyone wants to buy the same and write some drivers, it will be less work for me. It currently needs drivers for:

  • Audio (I have almost fixed that yesterday with the help of Korli)
  • Dual output graphics (I spent hours on my previous machine and eventually gave up, but the new one has the same problems, so I’m looking into it again)
  • WebCam (this is standard USB, so get any one standard USB webcam working, and most of them will work)
  • SD card reader (I wrote most of the driver, there are some fixes to do, here as well, once it’s done the fixes will benefit all other machines with standard SDHCI controllers)
  • Some things I don’t really need: IR scanner, fingerprint reader, smartcard reader, neural accelerator, … but I think I’d rather spend my time fixing other issues than working on 100% supporting all of this just to claim the machine is 100% supported.

This raises another question: nowadays, the hardware is fairly standardized. Many machines are very similar, and once one is supported, adding support for the other ones is rather easy and does not need as much time as people seem to think it does. That time is easily compensated when I decide to buy a new, faster machine, and I can immediately start using Haiku on it. Moreover, the drivers are well tested on a variety of hardware, meaning they are well written, and take into account several possible corner cases, making them more reliable. So, each new machine we add support for decreases the maintenance costs by making the drivers and the internal architecture better. And it also increases the number of developers who can potentially work on Haiku, because they do not need to start by buying a specific machine.

So, what looks like a simple idea to fix every problem, in fact does not fix anything, and in the long run, introduces more problems because things in the system will get hardcoded to a specific machine, making it harder to support any other one when we decide to do so.

6 Likes

May I assume … it was an example for the legendary

strictness of Draco ?. oh, pardonnez-moi , … DraKo ?

:cowboy_hat_face:

1 Like

This is not a practical concept for Haiku and the reasons why have been mentioned here on the forum again and again (and again). The threads should be quite easy to find using the search function, there’s really no need to iterate this discussion one more time.

1 Like

The issue is that hardware compatibility is a mystery for Haiku. Even I struggle with it and have been around the project for more than a decade.

We need something akin to Frizbe : Ultimate BeOS Hardware Compatibility Guide

Hello dcatt,

please have a look at hardware.besly.de.

Regards lorglas

2 Likes

That’s great and all but it ultimately needs to be discoverable on the project website. Maybe such a reference already exists and I have failed to discover it. Would be great if the project itself hosted such a list.

This first problem you mention is not necessarily an issue with all hardware that could be targeted. Raspberry Pi currently guarantees that their main models stay in production for 10 years. As an example while it seems very outdated by modern standards Raspberry Pi 2 will be discontinued only in 2026. Other devices like the Framework laptops might also be interesting in this regard.

2 Likes

But by the end of their life they are still outdated, even if still “supported” (what does that even mean in this context?)

Anyway, ycu can have a look at MorphOS or ReactOS to get an idea how a system supporting only one ora small set of machines is doing. I wculd say, not so well compared to Haiku, and moving to different machines when the time comes causes them great difficulty.

1 Like

They can still be bought as new because the hardware is still being produced and I guess maybe there could be firmware updates for fixes even late in the product’s life.

If I look at MorphOS or ReactOS I don’t necessarily see a huge difference to Haiku but I guess for people who want to run it it would be helpful to have a well working piece of hardware that anyone can easily acquire for a long period of time. Even if it is outdated, I doubt most people running Haiku are really shooting for the bleeding edge of computing power anyway.

First, please don’t tell other people what they “have” to do, it is rude.

Just choose a supported device, there are a LOT of supported devices

2 Likes

I run Haiku on my main computer, and yes, I ned ssed the “bleeding edge” to manntain WebKit which is a quite large project. Other people do video processing and all kind of other things that need a powerful computer. And… it just works. So, what would we win by restricting ourselves to some specific hardware, except hreventing people from running Haiku on machines they already own or letting them shop from all the available market?

1 Like

“Recommended device” is not necessary for x86 or x64 since compatibility between standard PCs are high.

However, for porting to other ISA, recommended/prioritized devices list can be provided for the developers looking for devices to develop their port to RV/ARM. If they are buying one for development, they can easily find out which system can they develop on with the most community support. You may develop on other devices, but there are less talented developers that can assist you.

Getting Haiku fully functional on one alternate ISA device (e.g. VF2) is also more newsworthy than only getting 30%-40% working on varieties of devices. What does newsworthy mean to the development? Getting VF2 and RV as an example, by an iconic breakthrough of fully functional port onto VF2, we can have effective promotion to RV developers’ communities and attract their developers who understands RV systems very well to join Haiku port development. They are also very eager to gain wider range of software support and cooperation between open source communities.

At the same time, for ARM port, there have been long lasting complaints on Rpi using proprietary chipsets, making development of drivers becomes difficult. Does switching to open source replacements of Rpi like Radxa ROCK or Banana Pi help?

Well, this conversation started off on the wrong foot, but I think it’s fair to say that there is an issue here. Even if it isn’t going to deliver us to Bill Gates or whatever.

Maybe there’s a way to get something going that will lead to reliable, relatively full support for a current production laptop. That would be pretty helpful, I think.

I am confused, and I’m sure the many other people already running Haiku on current production laptops also are.

Again, the few things that are not working are not specific to one model. We need one driver for i2c touchpads, for example. Once that works, it doesn’t work on just one machine, it works on all of them. And that’s the same for pretty much all other devices. Get any one UVC USB WebCam working, and all of them will work. Get one laptop with Intel graphics to display things on two monitors, and they all will. And so on.

7 Likes

“Reliable.” I’m not talking about making a bet on “supports Haiku” that people often win. Of course that’s a good thing, because as you’ve said, in a typical case it’s someone trying to run Haiku on a laptop they already have. Is the success rate pretty good? I hope so.

But when you want to get a laptop and be sure it will work? If there’s a solid answer for that, I think it would be a good thing for the front Haiku page. That isn’t going to be out of date a month after it’s posted. If there isn’t that solid of an answer, then that’s what I’m talking about – maybe there’s a way to get something like that going.

3 Likes

I don’t think there should or even could be a restriction that prevents anyone from doing that but I do think focusing resources on making sure one device that anyone can buy works really well could make some sense. Doesn’t mean nobody can work on any other hardware compatibility or that the system stops working on everything else…

1 Like

Pi4 was effectively unavailable for normal users for two or three years; and they’ve had multiple models with binary blob drivers so it isn’t a great target either.

Framework is nice but exceptionally expensive.