4 ms·

Metasepi Report: Writing NetBSD Sound Drivers in Haskell [pdf]

59 points by lelf 12y ago

nyoro 12y ago
Is it real?
- codygman 12y ago
  Yep. Looks like they are using ATS these days: http://www.metasepi.org/en/posts/2014-10-19-osc-tokyo.html http://www.metasepi.org/en/posts/2014-10-19-osc-tokyo.html
hardwaresofton 12y ago
Great work -- Learned something today about reentrancy, also wasn't aware of jhc as an alternative haskell compiler. Excited to see more haskell work in embedded systems
chocolateboy 12y ago
> Writing NetBSD Sound Drivers in Haskell http://www.blarworld.net/hackernewsgen.html http://www.blarworld.net/hackernewsgen.html
- codygman 12y ago
  Nope! I'm happy to say this one is real :)
akbiggs 12y ago
This was accessible and interesting, thanks for the link. A few questions: 1. Does the overhead of creating a new CLH every time a Haskell function is invoked from C introduce any performance concerns(speed didn't seem to be well-covered in section 6)? 2. How should Table 5 be interpreted? Is the main goal to show that B5 and B6 do not use the garbage collector if it's unnecessary for the task? Not sure I understand how the event-driven mutator fits into the situation.
- masterq 12y ago
  > 1. Does the overhead of creating a new CLH every time a Haskell function is invoked from C introduce any performance concerns(speed didn't seem to be well-covered in section 6)? Yes. It needs to initialize the heep at any function entry point. The time is proportional to number of all thunk on the Haskell application. However, in future design, we may cache the initialized heap structure. > 2. How should Table 5 be interpreted? Is the main goal to show that B5 and B6 do not use the garbage collector if it's unnecessary for the task? Not sure I understand how the event-driven mutator fits into the situation. Yes. Interrupt occur on Table 5. B5 and B6 are also using GC. However, GC trigger doesn't occur while the Haskell contexts are live, because the context's heap is big enough to escape first GC trigger watermark.
fithisux 12y ago
i am a bit confused. is jhc or ats2 the recommended way to write device drivers?
- masterq 12y ago
  Ah, sorry. Today, I think ATS2 is better than any Haskell compiler.
  - masterq 12y ago
    For kernel programming. < ATS2 is better

59 points by lelf 12y ago

nyoro 12y ago

Is it real?

codygman 12y ago
Yep. Looks like they are using ATS these days: http://www.metasepi.org/en/posts/2014-10-19-osc-tokyo.html http://www.metasepi.org/en/posts/2014-10-19-osc-tokyo.html

hardwaresofton 12y ago

Great work -- Learned something today about reentrancy, also wasn't aware of jhc as an alternative haskell compiler. Excited to see more haskell work in embedded systems

chocolateboy 12y ago

> Writing NetBSD Sound Drivers in Haskell http://www.blarworld.net/hackernewsgen.html http://www.blarworld.net/hackernewsgen.html

codygman 12y ago
Nope! I'm happy to say this one is real :)

akbiggs 12y ago

This was accessible and interesting, thanks for the link. A few questions: 1. Does the overhead of creating a new CLH every time a Haskell function is invoked from C introduce any performance concerns(speed didn't seem to be well-covered in section 6)? 2. How should Table 5 be interpreted? Is the main goal to show that B5 and B6 do not use the garbage collector if it's unnecessary for the task? Not sure I understand how the event-driven mutator fits into the situation.

masterq 12y ago
> 1. Does the overhead of creating a new CLH every time a Haskell function is invoked from C introduce any performance concerns(speed didn't seem to be well-covered in section 6)? Yes. It needs to initialize the heep at any function entry point. The time is proportional to number of all thunk on the Haskell application. However, in future design, we may cache the initialized heap structure. > 2. How should Table 5 be interpreted? Is the main goal to show that B5 and B6 do not use the garbage collector if it's unnecessary for the task? Not sure I understand how the event-driven mutator fits into the situation. Yes. Interrupt occur on Table 5. B5 and B6 are also using GC. However, GC trigger doesn't occur while the Haskell contexts are live, because the context's heap is big enough to escape first GC trigger watermark.

fithisux 12y ago

i am a bit confused. is jhc or ats2 the recommended way to write device drivers?

masterq 12y ago
Ah, sorry. Today, I think ATS2 is better than any Haskell compiler.
- masterq 12y ago
  For kernel programming. < ATS2 is better