7 ms·
Why they chose Next over Be is an interesting lesson itself. I still can't remember the technical merits of the Be operating system. Someone tried to get me to
by gwbas1c 9y ago
Why they chose Next over Be is an interesting lesson itself.
I still can't remember the technical merits of the Be operating system. Someone tried to get me to use it back in 1999, but there were no applications worth using.
- jhugg 9y agoBe was designed to be lightweight and parallel. It was optimized for many threads and the API pushed you to embrace threading. It also had pretty decent scheduling for media stuff. Not quite real-time OS, but closer. It was screaming fast and responsive. The catch is that it was much less mature. It was single user and had very little real security. It had basic Posix support, but NeXT has much more rich BSD-ishness.
- gwbas1c 9y agoAbout 9-10 months later I installed Windows 2000 and I got all that... And a rich ecosystem of applications!
- jghn 9y agoBy 1999 it was too late. They'd already transitioned to x86 and that was their death knell. At the time most hardware would provide drivers only for Windows which meant that other OSs needed to create their own. This sorta-kinda worked for things like Linux where the masses would crowdsource them, but for a commercial OS like BeOS they had to spend a ton of time working on this instead of the actual OS. My recollection at the time was as soon as they entered into this phase the forward progress on BeOS itself ground to a halt, and within a few years the mainstream OSs had started to at least pay lip service to its key features. Any momentum they had was completely lost.
- walrus01 9y agoBy late 1999 it was also too late because Microsoft had taken NT4.0 Workstation, put a much better GUI on it and made it even more stable and reliable, and released it as Windows 2000. It was pretty regular to see Windows 2000 workstation machines with uptimes of 25+ days, which was unheard of in the consumer Win95/Win98/Win98SE era. You could actually kill a misbehaving application and not risk crashing the whole OS.
- tarsinge 9y agoI remember my Windows 2000 Pentium III was a joy to use (despite being an heavy Mac user), so much that I never upgraded it to Windows XP.
- linguae 9y agoI concur. In my opinion, Windows 2000 was the best version of Windows. It had a very clean interface, and it was a very stable operating system. It also lacked the annoyances of later versions of Windows, such as activation, mandatory updates, and telemetry. It was everything I ever wanted from an operating system, and honestly, even though I'm a Mac user, I'd use an updated version of Windows 2000 if Microsoft ever decides to make one. I've been keeping an eye on the ReactOS project for this reason.
- pjmlp 9y agoLinux would have been dead by now if companies did not saw in it the opportunity to avoid paying for UNIX licenses. Or in the case of UNIX vendors, outsource their development costs.
- walrus01 9y agoWhy they chose NeXT over Be... Because the Unix nerds told them that if they wanted a rock solid workstation OS to replace legacy-MacOS (which was getting very long in the tooth at that time), it would be a good idea to have something BSD/NextSTEP based rather than whatever BeOS was. If you look at OSX v10.0 it's pretty obvious that it's NeXTSTEP with a MacOS-looking GUI slapped on top of it. They've been incrementally upgrading and developing it since then. BeOS was very interesting, but one thing it was not was anything close to a BSD/Unix platform.
- mattl 9y agoIf you look at Mac OS X 1.0 Server, it's even more obvious. Rhapsody is pretty interesting stuff. EDIT: Missed the link. http://rhapsodyos.org/ http://rhapsodyos.org/
- walrus01 9y agoyeah, the wikipedia page explains it in far better detail: https://en.wikipedia.org/wiki/Mac_OS_X_Server_1.0 https://en.wikipedia.org/wiki/Mac_OS_X_Server_1.0
- qubex 9y agoAt the time I was (coincidentally) using NeXTStep at work and a BeBox at home, and I have a bit of extra perspective to add. You are quite correct that NeXT was undoubtedly, upon close inspection, the ‘superior’ OS (UNIX lineage, stable, multi-user, etc.) but as a user (and sometime programmer) it wasn't that evident: my relatively cheap dual-604e BeBox outperformed my dual PentiumPro NeXT machine by a significant margin, or rather, felt like it did because the “pervasive multitasking” made the system extraordinarily responsive, and the lack of legacy code and backwards compatibility definitely kept the system snappy and lithe. Multi-user support didn't really change much from the user perspective either, as security was something of a remote concern and running any kind of ‘server’ on one's desktop was pretty exotic. BeOS was POSIX compliant to a degree such that many people still (erroneously) believe that it is a Unix or Unix-like, and you could run all the open-source command line tools you could possibly want. The BeFS filesystem's metadata capabilities were extraordinarily, and I liked both UIs equally (the etched vs. graffiti aesthetic legacy has etched itself into my neural circuitry), but probably NeXT was more innovative in terms of user interaction and rendering model whereas BeOS basically aped Windows and Classic MacOS quite closely. NeXT's development tools and environment were clearly superior. It was comparatively a resource hog. So yes, there was a technical edge, and probably knowledgeable influencers at the upper echelons of power probably recommended as such. But at the time I had the feeling that Jean-Louis Gassée (the founder and CEO of Be Inc.) significantly overplayed his hand, and made ridiculous demands because he thought he had Apple cornered by virtue of their failed OS development attempts, growing obsolescence, and perceived lack of serious alternatives. There's the diffuse feeling that Gil Amelio (Apple's then-CEO) eventually grew weary of batting back the demands and started looking for apparent alternatives just as a means of introducing some competition to the negotiations, and eventually the plonked for NeXT much to everybody's surprise. He even made a droll remark about “choosing Plan A rather than Plan Be” or something to that effect. All things considered, it was a very counterintuitive decision, mainly because NeXT had such hefty resource requirements and we hand't quite yet absorbed the implications of Moore's Law (or rather, were used to Microsoft squandering any and all increase in hardware performance with increasingly inefficient code, leading to no perceivable improvement year over year). Amelio deserves more credit for having seen through all of that and made a brave, surprising, and obvious-only-in-hindsight decision that favoured a mature, stable, and extensible platform. At the time I felt it was an utter mistake, but I was fourteen, and pretty ignorant of the ways of the world. I think history speaks for itself. Amelio deserves more respect for the decision he made, because it was ultimately his decision that led to adopting the superior, extensible technology, and of course brought Steve Jobs back on board (who promptly stabbed him in the back and ousted him in a boardroom coup). I had a brief flirtation with Windows NT4/2000/XP in 1999-2003, but then I switched to OSX 10.2 on a dual PowerMac G4 and have never looked back since. Only of late have I started to feel that as Apple's focus has switched to iOS on mobile devices they've started to let their development effort and quality control really slip to borderline-unacceptable levels. I can understand that OS X is no longer the enabler for high-margin computer sales it once was, and that that role has passed on to iOS, but OS X (macOS of late) is the platform where the apps that form the bulwark of the iOS ecosystem get developed. Apple really shouldn't let the desktop OS languish or endure the kind of high-profile mishaps that have occurred of late.
- Jtsummers 9y agoFantastic preemptive multithreading, especially for the time. The file system (BFS) was basically a database, with metadata capabilities that allowed applications to store their data in files and open up their data format (after a fashion) to any other application. Want an MP3 player? Don't need to open the files and get their ID3 tag info, it's in the file system. You can query it with "filetype:mp3 artist:bach" or whatever and you'd get a list, updated when other songs files were added or changed, of all mp3s meeting your requirement. Same for any other document. Contacts can be stored the same way, rather than hidden away inside an arbitrary file format you can store them with each bit of info as metadata (query on the phone number and find everyone associated with it), email, anything. The real benefit of this filesystem-as-database approach was that it breaks the data silo situation you find in many (most?) other OSes. Downsides were the dearth of applications, being single-user, and for a time being PPC only. Opening it to x86 opened it up to many more users, but then the OS became the product (instead of the BeBoxes they also sold) which was much lower price and profit. It was a POSIX-compatible OS (good), but not really a Unix in the way users of Unix would expect. This did make porting of applications a bit easier, though. They were also crippled by Microsoft's actions with Compaq (HP?), where MS threatened to change the favorable licensing costs for MS Windows if they sold BeOS on their computers.
- tech2 9y agoI was never too sure about the POSIX compatibility being a good thing. Sure, it meant that BeOS got more software more easily from other (Linux) sources. But it meant that people didn't write for the OO API, POSIX threading was worse, as were a lot of other things. Since things were then just ported to Be rather than written directly for it, it kinda felt like POSIX compat was Be in its death-throes. I bought in at R3/3.1, and I regretted it since as it was just a sad slow slide after that :(
- deadmetheny 9y agoBeOS had multithreading back when that was fairly new for consumer hardware. It was great for media purposes but the OS itself wasn't particularly well-developed at that point and Be wouldn't drop its overblown asking price, so Apple went with NeXT, and the rest is history. BeOS is a really neat little OS. I like it a lot, I often wonder how different things would be if they had taken that direction.
- hota_mazi 9y agoFor what it's worth, I think AmigaOS was superior to BeOS in pretty much every way, even as soon as the early 90s. Basically, AmigaOS 1992 was a more advanced operating system than BeOS 1997. Pity Commodore completely squandered their opportunity and Windows 95 killed them dead.
- qubex 9y agoAmigaOS was cooperatively multitasked whereas BeOS was preemptively multitasked, and I don’t think AmigaOS has memory protection (BeOS certainly did). AmigaOS didn’t support multiprocessing either, if I’m not mistaken. The BeFS filesystem was a pretty impressive marvel in terms of metadata and Amiga certainly lacked anything even remotely similar.
- bronson 9y agoBeOS did some tricks with the MMU to catch overruns but didn't have proper memory protection (in 1999 anyway, when I last used it). Any app could scribble on the kernel and hardware devices. This was part of why the graphics felt so flippin fast: appkit just wrote to the GPU & framebuffer itself. No need for any context switches.
- qubex 9y agoThat’s not what I remembered, but upon reflection you’re probably right since lack of multiuser support is difficult to reconcile with a true protected memory model. I’ve still got The BeOS Bible on my bookshelf so I’ll go have a read through the relevant chapters and take a trip down memory lane (ahem) just for nostalgia’s sake.
- phs2501 9y agoBe's use of C++ (at the kernel/library API/ABI level) would have doomed it in the long term IMHO. There have been several incompatible C++ ABI bumps since then, all of which would have broken compatibility. See the current Be clones which still have to use an ancient version of GCC to not break compatibility with existing software. NeXT's ABI interfaces (C and ObjC) were and are much more stable. (I'm not saying don't use C++, just don't use C++ and expect a stable binary API. You need to wrap it in a C API instead unfortunately. There's too much magic in C++ to have a stable ABI and still be able to evolve the language.)
- twoodfin 9y agoThis is a great point. I understand why a C++ ABI is so hard to keep nailed down, but the inability to do so has really hampered adoption of C++ as a true “system” language.
- rahoulb 9y agoLots of good answers here. All I will add is that Mac OSX was dog slow for years, from Rhapsody through till 10.2 (which I still thought was slow but was at least usable). BeOS running on awful hardware felt lightning fast and responsive in comparison. Of course, Apple without His Steveness would be an entirely different proposition