Comments on: Is Functional Abstraction Too Clever? http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/ Git, .NET and more Mon, 01 Apr 2013 17:24:00 +0000 hourly 1 http://wordpress.org/?v=3.4.2 By: Alexander Abramov http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-24 Alexander Abramov Fri, 30 Oct 2009 19:11:58 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-24 Nice post and comments, functional style needs to be brought to broader audience. It is is just a tool with its own uses, and a very expressive tool at that. Nice post and comments, functional style needs to be brought to broader audience. It is is just a tool with its own uses, and a very expressive tool at that.

]]> By: Keith Dahlby http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-23 Keith Dahlby Mon, 19 Oct 2009 16:27:01 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-23 @Chris ~ You're right except that after Append() we're sorting the full sequence so the 0 will end up first anyway - a slightly more optimized (but maybe less readable?) version would sort the random values first and then prepend/append as you said: .Take(slots - 1) .OrderBy(i => i) .Prepend(0).Append(max) ... @Chris ~

You’re right except that after Append() we’re sorting the full sequence so the 0 will end up first anyway – a slightly more optimized (but maybe less readable?) version would sort the random values first and then prepend/append as you said:

.Take(slots – 1)
.OrderBy(i => i)
.Prepend(0).Append(max)
…

]]> By: Chris Falter http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-22 Chris Falter Mon, 19 Oct 2009 15:45:35 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-22 Looks like there's a bug in calling Append(0, max)? Based on the implementation of Append, it likes like it concatenates 0 and max as the last 2 members of the list. That's what IEnumerable<TSource>(IEnumerable first, IEnumerable second) does--in your case, "0, max" being the IEnumerable second. Instead, I think you would have to do the following: .Prepend(0) .Append(max) // etc. Other than that minor bug, I love the idea and the code. Excellent post! Looks like there’s a bug in calling Append(0, max)? Based on the implementation of Append, it likes like it concatenates 0 and max as the last 2 members of the list. That’s what IEnumerable(IEnumerable first, IEnumerable second) does–in your case, “0, max” being the IEnumerable second.

Instead, I think you would have to do the following:

.Prepend(0)
.Append(max)
// etc.

Other than that minor bug, I love the idea and the code. Excellent post!

]]> By: Jeremy Gray http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-21 Jeremy Gray Mon, 19 Oct 2009 03:57:23 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-21 @Keith - For what it's worth, I double-checked Random.Next's behavior just now and I was in fact having a brain fart earlier today. Your impl is both clear in intent and clear of bugs. :) @Keith – For what it’s worth, I double-checked Random.Next’s behavior just now and I was in fact having a brain fart earlier today. Your impl is both clear in intent and clear of bugs. :)

]]> By: David Morton http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-20 David Morton Sun, 18 Oct 2009 01:46:51 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-20 Check the StackOverflow link. I've posted another possible solution that seems to work for the purposes, and still seems to be relatively fast. It's certainly inspired by the code on this post, however, but I've chosen instead to use a dictionary instead of creating a whole pairwise method, though Pairwise isn't a bad idea. Check the StackOverflow link. I’ve posted another possible solution that seems to work for the purposes, and still seems to be relatively fast. It’s certainly inspired by the code on this post, however, but I’ve chosen instead to use a dictionary instead of creating a whole pairwise method, though Pairwise isn’t a bad idea.

]]> By: Keith Dahlby http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-19 Keith Dahlby Sat, 17 Oct 2009 19:54:30 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-19 @Paul ~ There really isn't a clean way to represent Pairwise with the standard operators - the F# implementation is imperative as well. @Adam ~ I don't think corporate environments care about the language used, per se. They care about support and availability of developers. Until MS can prove that F# is ready for the big time (they're trying) and there are more than a token number of developers that really know F#, I don't see that happening. Eventually we'll get there, but in the meantime I would advocate all developers learning how to apply functional concepts in their current language of choice to ease the transition if/when that day comes. @Jeremy ~ I don't believe in bugs, only undocumented features. ;) My intent with Values is to provide an infinite sequence generated with Next(min,max), with terminating the sequence delegated to Take(While). I think it needs a better name than Values, but I can't think of one that I like. @All ~ I figured the perceived cleverness was due to its non-imperativeness, but sometimes it's hard to be objective about solutions that I rather like. Thanks for the feedback! @Paul ~ There really isn’t a clean way to represent Pairwise with the standard operators – the F# implementation is imperative as well.

@Adam ~ I don’t think corporate environments care about the language used, per se. They care about support and availability of developers. Until MS can prove that F# is ready for the big time (they’re trying) and there are more than a token number of developers that really know F#, I don’t see that happening. Eventually we’ll get there, but in the meantime I would advocate all developers learning how to apply functional concepts in their current language of choice to ease the transition if/when that day comes.

@Jeremy ~ I don’t believe in bugs, only undocumented features. ;) My intent with Values is to provide an infinite sequence generated with Next(min,max), with terminating the sequence delegated to Take(While). I think it needs a better name than Values, but I can’t think of one that I like.

@All ~ I figured the perceived cleverness was due to its non-imperativeness, but sometimes it’s hard to be objective about solutions that I rather like. Thanks for the feedback!

]]> By: Jeremy Gray http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-18 Jeremy Gray Sat, 17 Oct 2009 19:20:05 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-18 re: there being a bug - unless I'm having a Complete Saturday Brainfart about the behaviour of the Values extension method, that is. ;) re: there being a bug – unless I’m having a Complete Saturday Brainfart about the behaviour of the Values extension method, that is. ;)

]]> By: Jeremy Gray http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-17 Jeremy Gray Sat, 17 Oct 2009 18:59:56 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-17 This code is not too clever. In fact, it would be difficult to create a more direct representation of the stated requirements! The meaning of Pairwise might be the only part someone would find unfamiliar and even that part takes only a moment to understand. Code one step too clever obscures intent (and bugs) as much as the old-school procedural style code that was accepted over your version. The goal is to find the sweet spot, which I think you've done nicely. Had that accepted implementation contained a bug, it would almost surely be lost in the noise and harder to spot than the small bug I very quickly spotted in yours. :D The fact that I could spot it so quickly is, IMHO, a very good thing. Good code strives to make intent clear and bugs obvious. This code is not too clever. In fact, it would be difficult to create a more direct representation of the stated requirements! The meaning of Pairwise might be the only part someone would find unfamiliar and even that part takes only a moment to understand.

Code one step too clever obscures intent (and bugs) as much as the old-school procedural style code that was accepted over your version. The goal is to find the sweet spot, which I think you’ve done nicely.

Had that accepted implementation contained a bug, it would almost surely be lost in the noise and harder to spot than the small bug I very quickly spotted in yours. :D

The fact that I could spot it so quickly is, IMHO, a very good thing. Good code strives to make intent clear and bugs obvious.

]]> By: Stephan Schmidt http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-16 Stephan Schmidt Sat, 17 Oct 2009 15:51:35 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-16 Nice post. I'm not sure if all this functional programming stuff doesn't just boild down - in these examples on list processing - to a pipeline. I've solved many problems of this kind with an OO pipepline of Mappers and Filters. Oh and I like functional programming. Cheers Stephan http://www.codemonkeyism.com Nice post. I’m not sure if all this functional programming stuff doesn’t just boild down – in these examples on list processing – to a pipeline. I’ve solved many problems of this kind with an OO pipepline of Mappers and Filters.

Oh and I like functional programming.
Cheers
Stephan
http://www.codemonkeyism.com

]]> By: Adam Wolf http://lostechies.com/keithdahlby/2009/10/17/is-functional-abstraction-too-clever/#comment-15 Adam Wolf Sat, 17 Oct 2009 15:36:56 +0000 /blogs/dahlbyk/archive/2009/10/17/is-functional-abstraction-too-clever.aspx#comment-15 Too clever, No. This is a great example of how functional programming dissects problems and solves them. I can't imagine people who actually know how LINQ, Yield and extension methods would think this is too clever. The level of abstraction in the GetSlots method is perfect. I guess some people need to have everything in front of them to feel comfortable about the code but are willing to use the BCL without browsing the source. Does F# have a chance in corporate america? Too clever, No. This is a great example of how functional programming dissects problems and solves them. I can’t imagine people who actually know how LINQ, Yield and extension methods would think this is too clever.

The level of abstraction in the GetSlots method is perfect. I guess some people need to have everything in front of them to feel comfortable about the code but are willing to use the BCL without browsing the source.

Does F# have a chance in corporate america?

]]>