I loved BeOS. There was a period of about a year where it was the only OS I ran. I bought multiple versions of the OS, even the Professional Edition of BeOS 5 that I could’ve gotten more-or-less for free to give them some money.
I love Haiku. I love what they’re trying to do.
That being said…Haiku fell prey to the classic second-system effect. What started as an attempt to clone BeOS 5 (which was a truly small operating system) now has:
The package filesystem (which is the most elegant package management system I’ve ever seen, but is still a lot of code and necessitated a lot of work for the bootloader, etc).
Remote drawing support in app_server
HTML5 targets in the app_server
Network block devices
Support for tons of different file systems
A systemd-like initialization system
…
I would’ve loved to have seen Haiku simply target BeOS 5 source-level compatibility and then release Haiku 1.0. All of that stuff could’ve come later. Instead it’s been fifteen years of putting lots of half-baked features on top of what used to be a small operating system.
Note that this sounded harsher than I meant it to. Haiku is awesome and I wish it could do what I needed it to do for my day-to-day work (I have to be able to run a lot of VMs and use Google Hangouts videoconferencing). The people working on it have done great things. I just am saddened that you can’t really describe Haiku as “small” any more, at least not when you look at the code.
I don’t disagree, but I don’t think second-system effect is what killed them as such–and I don’t think they’re dead.
The first thing that “killed” them, but that I don’t know how you would’ve avoided at the time, was a commitment to binary backwards compatibility. That forced a lot of decisions on them that wouldn’t have applied if they’d aimed for source compatibility instead, ranging from locking in compiler versions to needing to emulate bugs to everything else. It was the WINE project, but also had to worry about a kernel and a file system and more. That’s just honestly a lot of minutia to bite off.
The second thing, and this is complicated, is that BeOS really wasn’t designed for today’s world. It was a single-user system with no meaningful security provisions, and its POSIX support was okayish but not amazing at the time. In a world where Haiku had just nailed things from day one, they’d have immediately found themselves in a position where they needed to move forward, anyway–and this is a key area where even modern-day Haiku isn’t quite sure what to do.
The third thing is that programming for BeOS…I mean, it’s fine, but it was behind mainstream options for both Windows and OPENSTEP at the time. It was nice compared to Mac OS 8, but by the time Haiku launched, that was already not the target. But that in turn meant that getting people to develop for fun was tricky.
I’m not saying they haven’t gotten horribly distracted (they have), but I’m not sure that I agree that their distraction killed them. In fact, the way they’ve moved forward, it may’ve saved them in a way. From where I sit, they took a step back, said, “okay, let’s still aim for compat, but we’re gonna also make this be interesting to devs by doing all these other things,” and that second part is what’s kept them alive.
I may be wrong, and that’s fine. I definitely don’t use Haiku as my daily driver, and I doubt I ever will, and someone who had been hoping for that would probably have a radically different takeaway. But because of the fascinating stuff they’ve gotten distracted by, I keep coming back to play, look, and (on very rare occasions) contribute, similar to how I love following Dolphin.
And I think the dev community’s big enough for that. That can be a route to success, not failure. It just requires acknowledging that Haiku’s goal has, after well over a decade, changed, and accepting that.
They basically have hit R5 level compatibility - the prereleases are fine for that if all you wanted was that. Gobe Productive might be a little wonky, but it does work.
There was an announcement I don’t have bookmarked where a project member said they no longer had hope for it being a full-fledged OS that will go mainstream. The goal was more about experimentation and fun. Too bad, though, as I’m with you in wishing they just rolled out everything that was good in BeOS as a solid, paid or free release. Then, they can go from there. Might have made some money off kiosks, infotainment, or appliances for app servers where concurrency helped. There were possibilities.
Good example of it working is MorphOS team that was constantly rolling out updates to a beautiful Amiga they sold for $100-200 range.
I wrote a ton of applications for it. I wish Haiku had pushed forward from where BeOS was.
When I first saw Haiku years ago, I was hopeful until I saw their goals. They were never goals that was goiing to lead to a usable forward looking OS that built on what made BeOS awesome.
I still have my 2 BeBoxes and can fire them up from time to time for fun and nostalgia. Sadly, it’s still the best, most efficient OS that I’ve used on a daily basis.
I loved BeOS. There was a period of about a year where it was the only OS I ran. I bought multiple versions of the OS, even the Professional Edition of BeOS 5 that I could’ve gotten more-or-less for free to give them some money.
I love Haiku. I love what they’re trying to do.
That being said…Haiku fell prey to the classic second-system effect. What started as an attempt to clone BeOS 5 (which was a truly small operating system) now has:
I would’ve loved to have seen Haiku simply target BeOS 5 source-level compatibility and then release Haiku 1.0. All of that stuff could’ve come later. Instead it’s been fifteen years of putting lots of half-baked features on top of what used to be a small operating system.
Note that this sounded harsher than I meant it to. Haiku is awesome and I wish it could do what I needed it to do for my day-to-day work (I have to be able to run a lot of VMs and use Google Hangouts videoconferencing). The people working on it have done great things. I just am saddened that you can’t really describe Haiku as “small” any more, at least not when you look at the code.
I don’t disagree, but I don’t think second-system effect is what killed them as such–and I don’t think they’re dead.
The first thing that “killed” them, but that I don’t know how you would’ve avoided at the time, was a commitment to binary backwards compatibility. That forced a lot of decisions on them that wouldn’t have applied if they’d aimed for source compatibility instead, ranging from locking in compiler versions to needing to emulate bugs to everything else. It was the WINE project, but also had to worry about a kernel and a file system and more. That’s just honestly a lot of minutia to bite off.
The second thing, and this is complicated, is that BeOS really wasn’t designed for today’s world. It was a single-user system with no meaningful security provisions, and its POSIX support was okayish but not amazing at the time. In a world where Haiku had just nailed things from day one, they’d have immediately found themselves in a position where they needed to move forward, anyway–and this is a key area where even modern-day Haiku isn’t quite sure what to do.
The third thing is that programming for BeOS…I mean, it’s fine, but it was behind mainstream options for both Windows and OPENSTEP at the time. It was nice compared to Mac OS 8, but by the time Haiku launched, that was already not the target. But that in turn meant that getting people to develop for fun was tricky.
I’m not saying they haven’t gotten horribly distracted (they have), but I’m not sure that I agree that their distraction killed them. In fact, the way they’ve moved forward, it may’ve saved them in a way. From where I sit, they took a step back, said, “okay, let’s still aim for compat, but we’re gonna also make this be interesting to devs by doing all these other things,” and that second part is what’s kept them alive.
I may be wrong, and that’s fine. I definitely don’t use Haiku as my daily driver, and I doubt I ever will, and someone who had been hoping for that would probably have a radically different takeaway. But because of the fascinating stuff they’ve gotten distracted by, I keep coming back to play, look, and (on very rare occasions) contribute, similar to how I love following Dolphin.
And I think the dev community’s big enough for that. That can be a route to success, not failure. It just requires acknowledging that Haiku’s goal has, after well over a decade, changed, and accepting that.
They basically have hit R5 level compatibility - the prereleases are fine for that if all you wanted was that. Gobe Productive might be a little wonky, but it does work.
There was an announcement I don’t have bookmarked where a project member said they no longer had hope for it being a full-fledged OS that will go mainstream. The goal was more about experimentation and fun. Too bad, though, as I’m with you in wishing they just rolled out everything that was good in BeOS as a solid, paid or free release. Then, they can go from there. Might have made some money off kiosks, infotainment, or appliances for app servers where concurrency helped. There were possibilities.
Good example of it working is MorphOS team that was constantly rolling out updates to a beautiful Amiga they sold for $100-200 range.
I used BeOS as my every day OS for 4 to 5 years.
I wrote a ton of applications for it. I wish Haiku had pushed forward from where BeOS was.
When I first saw Haiku years ago, I was hopeful until I saw their goals. They were never goals that was goiing to lead to a usable forward looking OS that built on what made BeOS awesome.
I still have my 2 BeBoxes and can fire them up from time to time for fun and nostalgia. Sadly, it’s still the best, most efficient OS that I’ve used on a daily basis.