I check back in with Haiku every year or so to see how it’s progressing. I always loved the concept but there was always something which made it impossible for me to use as a daily driver.
Well, that time is over now. I have a stable install, no problems with the graphics drivers, WiFi, an office suite, emacs, a browser that plays YouTube videos, a C++ compiler and Dosbox for games. The only thing I’m missing is a Common Lisp implementation and I can live without that (or try and port one myself).
You’ve done it guys, you’ve finally done it. Haiku is here (for me at least). Thank you so much.
I have said multiple times until now that Otter browser works good enough for all the sites I care for, some of them really modern and heavy. If you haven’t tried it yet please check it, it’s in the Depot.
Very glad that things work for you in Haiku as well!
Regarding Common Lisp you already got an answer. But since it’s close enough, I am glad to share that Clojure runs in Haiku! I was able to run Leiningen script and test that compiler works. Also I was able to install Clojure related add-ons to emacs. Since all it needs is the JRE I suppose it’s not a surprise. This is because I wanted to learn Clojure for a couple of years and I just began studying it. Always wanted to learn some lispy/functional thing to balance so many years with C++ and open my horizon a little bit
I love Haiku os … Some bugs sometimes but it works pretty well. Quite intuitive as a
Windows system more than Linux.
But too bad there is no GPU acceleration yet. That’s all I want to make it a better OS, a real workstation. I prefer it to Linux but it lacks hardware compatibility to get all its power.
Yes there are bugs on all OS but for navigating on world wide web by example through browser on Haiku that give me somes bugs…It’s my experiance of this. Webpositive or Otto browser seem difficulty rendering few pages about memory management. Other bug the wifi drivers for my card freeze the OS. That don’t come with other OS more mature about compatibility. I’m sure these bugs will be disolved with time but they still exists, not be blind.
Not sure what you mean. ECL ought to build on either just fine it has very limited requirements for building. Often the recipies don’t build on the other arch because it was never tested etc… so you can’t go by that.
Would the product owner or company really do something for Haiku? They might by a small chance for Linux because it has a quite large userbase compared to Haiku but Linux, low compared to Windows. Haiku is in a catch 22: it needs more users to get more notice from companies but it cannot get enough users for this because of not enough hardware driver support from companies.
How might this be solved: port hardware drivers from BSD or Linux, attract harware driver developers with financial support and generous donations from relatively small userbase, investigate if reverse engineering harware to create drivers could be speeded up with the help of AI; maybe genetic algorthms or type of AI (some of it not true AI) that is employed to help discover new drugs. Maybe IBM Watson type. I am not certain if deep learning AI would do much because it seems to take forever when applied to autonomous vehicles although eventually having reasonable functionality. Whatever is done it needs some very bold and inovative steps to get Haiku into the manstream where it would be really noticed.
For small userbase to make large donations, that userbase needs to be assured that someting would happen good with their donations in reasonable time.
Isn’t it possible to port the Linux / Unix or Beos drivers to Haiku? The drivers used in Linux, aren’t they most open-source?
Also I saw that VirtualBox is an OpenSource software offers graphics acceleration on certain guest OS.
Maybe collaborate with this virtual machine to develop drivers that communicate well between Virtualbox “graphics card” and Haiku in order to offer a preview of hardware acceleration under Haiku to users. Also VirtualBox include the VMWare graphics card apparently.
I checked it (Otter), and it works better than I expected. The test for me is Google Maps, which mostly works - it displays, zooms, pans. I’d love to be able to use my mouse buttons, though. To drag, for example, or mouse button combined with modifier keys, without which much of the functionality is out of reach.
For example, with WebPositive, invokes a menu of options related to the location. Otter, nothing. Or I try to drag the little homunculus to a location, to see if the street view works … but it seems there’s no dragging.
They are open source but under the GPL license, which we prefer to avoid. Also the code tend to be quite messy and without much comments in Linux, and they refuse to have a stable interface between the kernel and the drivers, which means keeping something like that working is a LOT of work, as Linux keeps changing its APIs all the time.
Here is a story about a research lab at my univeristy. They were working on FPGA routing algorithms and were considering something like this: let’s have an AI try many combinations on the real hardware and try to converge to something that behaves as we want. And the AI got amazing results, several % more efficient than state-of-the art algorithms would achieve. But they wanted to know how it achieved that. And when they looked closer, it turned out it exploited quantum field effects and many other undocumented behaviors of the hardware to get things done. Which meant it would not behave in a stable way, it would be affected by temperatres, input voltages, etc. The idea was, of course, scrapped.
AI generating code will also never generate comments or documentation. And it’s already hard enough to reverse engineer things done by humans, so I don’t see how that would be in any way helpful.
No, Haiku needs more users who actively ask for support. There is indeed one aspect which is getting more users at all, but there is the other which is telling the companies that such users want to use their hardware and would buy it if there was support from the company. Both aspects are essential, and one of them can already be done by the existing users.
If you know some hardware driver developer who works this way, tell them about Haiku!
In my case I already have a stable full-time job and I find that this allows me to work on Haiku in my free time without too much worries (I know I can pay my rent, etc at the end of the month in a long term way). This ends up being better in my case than leaving that job and working on relatively short-term contracts full-time with Haiku (I did the longest one which was 1 year back in 2014).
The situation is different for everyone of course, and I’d love to see Haiku inc funding more contracts like this.
For now, ECL recipe doesn’t build because of strange unresolved dependencies. All them are here, but ECL build process defines some variables, which are used in wired way to result in wrong path (?) so it doesn’t find the necessary code. This problem seems to be resolvable by someone with right skills and sufficiently motivated.
I personally get Common Lisp run on Haiku with ABCL, which also needs Java runtime. Clojure is not Common Lisp.