Hacker News Comments on
From Haskell to Hardware
begriffs.com
·
125
HN points
·
0
HN comments
- This course is unranked · view top recommended courses
Hacker News Stories and Comments
All the comments and stories posted to Hacker News that reference this video.⬐ snakyThere's a good set of slides by Conal about the project⬐ eatonphilDid not watch the video, is this about Bluespec?⬐ None⬐ DevPadNone⬐ tincoNo, and it isn't about Clash either. It's a system they designed for a now defunct hardware startup called 'Tabula'.Apparently the technology was some sort of space-tech FPGA that could reconfigure at 2Ghz (i.e. close to the general max clock speed), they use Haskell's typesystem and non-sequentiality to build these circuits in a nice parallel way.
⬐ listicSounds very rad. I wonder why they closed and what they managed to accomplish?⬐ mng2The comments on this EETimes article[0] are snarky but the criticisms sound cogent to me.This posting convinced me to learn Haskell. But it requires to be more than just a programmer.Why I love Python, - because of scientific-friendly community.
Haskell is more than that: more algos, data structures, functional programming, math, - completely next level.
⬐ SixSigma⬐ agumonkeyWirth told us in 1976 that :Algorithms + Data Stuctures = Programs
https://en.wikipedia.org/wiki/Algorithms_%2B_Data_Structures...
⬐ vezzy-fnordNot sure what "more algos", "data structures" and "math" are supposed to mean here.⬐ jimmaswellThey're just different paradigms, one's not on a "higher" or "lower" level. They're good for different things.⬐ DevPad⬐ KenjiYeah, what I feel like it could be with Haskell: get a glass of good wine and solve interesting programming problem during evening.And write some beautiful code.
Be prepared to be much less productive in Haskell; you can't just get into it and start writing code. Everything must be well thought out before you start to write a line (and yes, that's how it's supposed to be but sometimes it's easier to just hack something together to see what issues might arise, and then make it beautiful). On the other hand, the beauty of Haskell is that the code is much cleaner and modular. Don't be fooled, though: Even in Haskell you can write very ugly code.⬐ dllthomas"Be prepared to be much less productive in Haskell[.]"Likely to be the case, but mostly because there's a lot to learn before you can be effective as an intermediate or senior level Haskeller.
"[Y]ou can't just get into it and start writing code. Everything must be well thought out before you start to write a line (and yes, that's how it's supposed to be but sometimes it's easier to just hack something together to see what issues might arise, and then make it beautiful)."
This is not the case at all. You can write Haskell that way, but remember that 1) with type inference, you don't need to specify the types of most things up front; and 2) you can change your types as you go. You can totally just start coding, and then later peel out this tuple into a named type and such. I find I'm most productive somewhere between the two extremes - start from the few types that I know must be fixed (by the fact that I'm interfacing with other code expecting them), then continue sketching in both types and code and let each guide the other.
"On the other hand, the beauty of Haskell is that the code is much cleaner and modular."
When you're doing it right, absolutely.
"Don't be fooled, though: Even in Haskell you can write very ugly code."
Also very much the case - I've done it!
⬐ merijnv⬐ psibi> "Don't be fooled, though: Even in Haskell you can write very ugly code."> Also very much the case - I've done it!
On the bright side, refactoring ugly code into not ugly code is substantially easier in Haskell than many other languages! ;)
⬐ dllthomasIt can be. It depends on the reasons your code is ugly. Meaningful refactors are much easier in Haskell. Small, local improvements in clarity aren't going to be substantially different between Haskell and other languages.Also, it's worth noting that Haskell has a couple ways it can get ugly that are less of an issue in other languages.
First, point-free programming can be incredibly clear when you've got a simple series of transformations of relatively simple data. Sometimes it winds up over-applied, in situations where it would be clearer to break things apart.
Second, most languages enforce a distinction between expression and statement. That Haskell doesn't gives more flexibility about how one breaks up expressions. This flexibility means that sometimes ideas can be expressed more clearly, but it's an additional thing to learn on the path to writing readable Haskell.
> Everything must be well thought out before you start to write a lineNot exactly. When you are initially learning the language, you have to think a little more. But once you are comfortable with it, the type signatures dictate you what to do. Also, it's quite easy to refactor to a different design because of the types.
⬐ codygman> Be prepared to be much less productive in Haskell; you can't just get into it and start writing code. Everything must be well thought out before you start to write a lineSomething about Haskell makes people enforce this upon themselves. You most certainly can hack things together quickly with Haskell after learning as much as you would in an imperative langauge so long as you can accept it will be ugly and unmaintainable, but that's not much different than other languages.
⬐ tome⬐ agumonkeyYeah, I don't recognise that restriction either.Should we read 'much less productive' as 'much more precise and solid, less buggy and forcede to be well thought through' ? Sometimes it feels that many systems confounds shipping lines with productivity.⬐ DevPadVery true. It's possible to write ugly code in any programming language, just so easy.⬐ creichertBe prepared to be much less productive in Haskell; you can't just get into it and start writing code. Everything must be well thought out before you start to write a lineThis is only true when you are becoming familiar with the language (arguably any programming language). Once you're able to get a grasp on how to program in Haskell, it's very possible to hack programs together quickly and refactor with confidence later.
I didn't even realized it was Conal Elliott. You're all invited (one at a time) to visit his blog and enjoy the ride. http://conal.net/blog/Especially if you're into Functional Reactive Programming, rendering, etc, ...
His [re]definition of 3d rendering was gold: http://conal.net/blog/posts/3d-rendering-as-functional-react...