Haiku’s monthly activity report for March is out has been out for weeks now, and it contains some interesting nuggets as the team moves closer to beta, but one stood out to me:
Kalisti5 got the PowerPC build working again. It is still not possible to boot PowerPC images very far, but at least it is now possible to compile them, and our buildbots are now happily doing so.
I find it interesting that there’s people at Haiku still working on PowerPC support. It’d be interesting if they ever manage to support Apple PowerPC hardware, if only to offer yet another choice besides MorphOS.
Either that or maybe they want to run on one of the various Amiga boards? Dunno. I don’t see much point otherwise.
There’s a lot more Power Macs than there are PPC Amigas….
@ galvanash
They have ported Haiku to the Sam460Ex by Acube systems: (aka AmigaOne 500, much cheaper than the high powered A-EON AmigaOne X5000 PPC boards)
https://www.youtube.com/watch?v=Ldh3foPCJJw
BTW, a bootloader was once leaked to run AmigaOS4.x on PowerPC Mac Minis, but no official support for that though.
And IIRC I stumbled once on some info about running OS4 on Mac Mini …wasn’t it without sound?
Edited 2018-04-27 00:15 UTC
And Linux?
And OpenBSD and NetBSD and FreebSD.
And, you know, OS X and Mac OS 9
I think that OpenBSD has worked better on old G5 hardware than Linux has. The OpenBSD policy of never cross-compiling releases and actually testing all packages tends to result in much, much better stability and usability than how most Linux distros ‘support’ less common hardware.
Whats the point? Seems like a waste of effort. Surely there are a better use of the teams limited resources?
Look at it this way. Do you think the person who spends their time working on a PowerPC port would be just as excited to work on fixing something like the HTTP2 parser? Or would they go off and work on the PowerPC port of NetBSD or some other OS? Or work on the port of clang or gcc, etc?
Large parts of projects like this exist because the developers want to work those portions of the project.
The truth is they do it just to personally annoy you.
That’s true!
Edited 2018-04-26 15:25 UTC
Your thinking in terms of normal “corporate” development teams. Most open source teams are nothing like that. Team members want to be there, because there is usually some particular thing they care about working on – and that is the thing they work on. That is quite often a particular piece of hardware. No one is making anyone do anything, and project direction is generally decided by consensus and willpower…
I don’t know personally, but I would not be at all surprised if the PowerPC version exists literally because one or two people wanted to do it for their own reasons, and they are the only people even working on it.
I do have some experience with the XBMC project (now called Kodi), and I can tell you that the original iOS port was done by pretty much one guy (davilla), and mostly because he just wanted to. Other devs got involved eventually, but for a long time his personal build tree was pretty much the defacto tree for that platform.
I was not a team member, I only did a few patches here and there. My own personal contributions were limited to the xbox platform, because that is what I ran it on. I did a few things for no other reason than I wanted them for my own use. If it worked and wasn’t stupid the team was very open to taking your patches. They still are afaict.
Edited 2018-04-25 20:02 UTC
Honestly, it was just about keeping all of our architecture ports working. (We already have quite a bit of PowerPC code + assembly in-tree)
PowerPC has been *completely* broken since we moved over to package management a few years ago (it needed bootstrapped). Fixing our PowerPC builds was a low hanging fruit which represents one more “building” architecture which could attract developers.
I’m personally more focused on ARM and RISC-V. Our main focus for R1 is still x86_gcc2 (BeOS ABI) and x86_64.
Edited 2018-04-26 15:28 UTC
What kind of ARM? Raspberry Pi?
I don’t have a need for it but all the small wins like this are an achievement with a sense of satisfaction. Good for them. For me, if they had a modern browser and Thunderbird (or equivilent) I’d use it for work and log bugs as I see them. All my work is server side so it’s low risk even if the while system goes belly up on me. Sadly I can’t contribute any other way, or I would.
Agreed, and I don’t even need Thunderbird, just a stable, modern browser with support for Firefox plugins. 95% of what I do on a computer at home is in the browser and the terminal, and the other 5% is when I boot into Windows to play games. Haiku already has a perfectly serviceable terminal, but the OS locks up every few hours on my old HP workstation dedicated to that OS. Once it’s stable enough I’ll stay on that machine as much as possible.
Oh our simple needs! I could probably get away with just a browser I guess. I’m thinking a web browser is the hardest piece of software to port and maintain on a niche os (and it’s dependent libraries).
Going back to PowerPC would complete a journey in many ways. The BeOS obviously came to notoriety on PowerPC hardware and popularity on the transition to intel.
But I doubt this opens us up to much old software, so that would be a real shame. The old BeOS used a exe format from the Classic era MacOS (PEF) and unless Haiku changed their policy, none of it will work on the Haiku releases for PowerPC.
If this runs on Old World hardware, I’ll give it a go at some point, but I down’t own any new world hardware, only Intel Macs, an old Wallstreet powerbook and 9500MP that had/had a copy or R5.03 on in last time it was booted, but has been in storage for a few years now..
Just installed BeOS on a 9600MP this morning. The real question is indeed whether applications and drivers will be ported to whatever Haiku cares to support. Would be nice if the devs have managed to include New World hardware, too.
There wasn’t a wealth of PPC Be software compared with Intel software back in the day, anyway.
Any news on the 68k version?