what went wrong?
Linux happened
Given how shitty driver supper linux-libre has I can only imagine how shitty GNU Hurd must be
>userspace kernel
>message passing
What more needs to go wrong before Stallman realizes it's inherently shit?
>>61543587
>>61543587
The giant overbloated microkernel from what I've read.
Hurd? More like TURD.
a unix-like microkernel is a poor design decision. it will always end up very slow because microkernels rely so much on ipc, and unix ipc has high overhead.
>>61543625
I have hurd installed in a VM
>no x64
>no SMP/multicore/multiprocessor
>no USB
>no sound, no sound system
>>61543587
Functional programming isn't sufficiently well explored yet to pull microkernels off.
A microkernel is de facto incredibly hard to debug.
Combine this with the technology of the time and the fact that GNU was ported to Linux a year thereafter, and it's no surprise that HURD became what it became.
The impetus to make a fully functioning kernel when you already have one is practically nix, which is why it became a research project.
With that in mind, most kernel developers "break their teeth on HURD", so it's not like it was a complete waste of resources. The merits of HURD are just more passive. HURD created a whole generation of kernel hackers that wouldn't have otherwise been there weren't it for the fact that there was a need for a fully free operating system.
>>61544189
Good one!
How will they ever recover?
>>61544827
>>61544904
It's weird how blatantly impractical functional programming is compared to how useful it is in a research environment. I mean, besides emacs, there's very little proof that functional programming can do anything useful, yet, basically every other language has benefited from research into functional programming. HURD is, in a way, incredibly representative of that.
>>61544904
Linus Torvalds doesn't like debuggers.
http://web.mit.edu/ghudson/info/linux-not-bazaar
>>61544827
>>61544936
functional programming has nothing to do with microkernels
>>61544827
>>61544936
>>61545014
Would it be even possible to create an OS in a purely functional language?
Since functional programming is all about having no state and immutable data and a Turing machine has by definition state and mutates data by reading instruction doesn't this make functional programming completely inappropriate for a low level OS programming?
Solution in search of a problem.
>>61543828
specifically so he may lay claim to any systems created atop it through argument with a lines of code metric