- 09:30 There’s a .travel TLD? #
- 13:38 Be’s Unix layer/network stack was always halfassed, even in Dano, and I think Apple would have lost lots of Unix nerds that way… #
- 13:40 Pervasive threading was way ahead of it’s time, and didn’t (really) matter until multicore procs were standard kit. #
- 23:44 coco modified my avatar: tinyurl.com/6pg5nx #
Automatically shipped by LoudTwitter
i was thinking the a while back about the threading and how Be kept predicting that we’d eventually hit the wall and instead have to move to multi-processor/multi-core systems. 10 years too early :)
i think they could’ve fixed the networking, as i think the core stack was probably fine, especially after the Dano improvements, but there were quite a lot of other API shortcomings that always made it a bitch porting applications and such, such as the threading model. wasn’t the network stack a zero-copy one in Dano?
They definitely could have fixed the network stack, but they never seemed to really have their heart in it…
Dano had a kernel space zero copy stack, but it was really flaky and, of course, perhaps this was caused by a small handful of bugs that, when fixed, would have turned it into a rock solid beast… But things like this made Be feel, at times, like a really cool proof-of-concept as opposed to a more substantial completed OS.
Of course, It’s easy to say all this a decade later.. I’ve gained the perspective of time but lost the perspective of the immediate. OSX wasn’t really ready for public consumption until 2002 when 10.2 was released, and Apple bought NeXT in 1996.. 6 years is a long while for a well-funded company to improve a piece of software, so who knows what the final Apple-ified Be would have looked like..