The Soft Side of Software

http://queue.acm.org/detail.cfm?id=3092954

by k4rtik • 9 years ago

102 25 9 years ago

queue.acm.org

zengid2• 9 years ago

I know it's easy (and fun) to hate on Microsoft, but damn do they employ some great engineers/researchers: Erik Meijer, Anders Hejlsberg, Simon Peyton Jones, etc..

redtuesday• 9 years ago

Yeah, when Leslie Lamport was mentioned a few days ago I thought to myself that MS has many, many more employees that I know from reading/watching in videos compared to apple, google, ibm etc.

nickpsecurity2• 9 years ago

It's Microsoft Research, specifically, that pulls in most of this great talent. Let's not forget the VMS's people run by Cutler that made the kernel, the Xbox hypervisor, and some other stuff. Add Butler Lampson if we're talking CompSci people. The teams behind Dafny, VerveOS, Ironclad, VCC, etc are way ahead of most language-based safety or formal verification in terms of cost/benefit analysis:

https://en.wikipedia.org/wiki/Butler_Lampson

http://rise4fun.com/

cwyers2• 9 years ago

I don't disagree with this, but Meijer left Microsoft a few years ago, I think he's at Facebook now.

interfixus2• 9 years ago

Hejlsberg did amazing work on Turbo Pascal/Delphi, which were basically his personal inventions. He then crossed over, and got himself lost in the C#/.net universe - technology some of us wouldn't touch with the proverbial ten foot pole, or one of any length, for that matter.

Considering that his (and my) compatriots Stroustrup and Lerdorf did C++ and PHP repsectively, a national apology might sort of be in order.

ktRolster• 9 years ago

He also did solid work in the C#/.net world.

UK-AL• 9 years ago

He wasn't just involved in c#, he was the lead on c#.

lindbergh• 9 years ago

Erik Meijer a very engaging person! If I recall correctly, he spearheaded the Linq project for C# back in mid-2000s.

In particular, I suggest his Haskell lectures to anyone interesting by theoretical aspects of functional programming. https://channel9.msdn.com/Series/C9-Lectures-Erik-Meijer-Fun...

chubot• 9 years ago

I just read over "Confessions of a used programming language salesman" [1] again.

I didn't quite get it the first time around, but now I get it. Good read. It is about the road from research to production, culminating in LINQ.

[1] https://scholar.google.com/scholar?cluster=17368623574878046...

zengid2• 9 years ago

Can anyone help me interpret the type signature Meijer used in his programming question? "Given a generic type:

    Cont r a = (a -> r) -> r

Prove that this type forms a monad."

I'm reading it as 'Container' passed 'r' and 'a' performs 'a' to 'r' which returns 'r'. (I'm guessing on verbs here)..

I've never come across a good way to read these type signatures.

bjl• 9 years ago

I read this as 'A cont of type r a is a function that, given a function of type a to r, returns an r'.

hdhzy• 9 years ago

Cont refers to the continuation monad. For details see http://www.haskellforall.com/2012/12/the-continuation-monad....

wz1000• 9 years ago

Cont stands for Continuation. A continuation basically represents a 'suspended' computation with an intermediate result of type 'a' and final result of type 'r'

    type Cont r a = (a -> r) -> r

Cont is a type constructor that takes two type arguments, r and a. This means that Cont r a can always be substituted by (a -> r) -> r. For example, Cont String Int is equivalent to (Int -> String) -> String

(a -> r) is the type of a function from a to r. For example, Int -> Bool is the type of a function from Int to Bool. (a -> r) -> r is the type of a function that takes a function from (a -> r) as its argument and returns an r. So Cont String Int takes a function from Int to String as its argument and finally returns a String.

http://www.haskellforall.com/2012/12/the-continuation-monad.... https://begriffs.com/posts/2015-06-03-haskell-continuations....

PopsiclePete• 9 years ago

My eyes glazed over as they usually do when trying to grok Haskell, but is that somehow related to partial computation and partial functions?

danidiaz• 9 years ago

Cont is one of the trickier monads to "get".

Cont r a is a computation that knows how to produce an "a" but, instead of returning it directly, passes it to a function that takes the "a" and produces an "r".

Compared to typical sequential execution, this gives more power to the "current" phase of the computation, because it might choose to invoke the passed a -> r function more than one time, none at all, inspect the resulting "r" and change course based on the result, etc.

chriswarbo• 9 years ago

I'm reading it as 'Container' passed 'r' and 'a' performs 'a' to 'r' which returns 'r'. (I'm guessing on verbs here).

This is an equation; it's a relating/defining one thing in terms of other things. In particular, there's no concept of "performing" (e.g. there's no time, state, etc. here)

In this case "Cont" is just a name; we can tell that, since the equation tells us that "Cont r a" is equal to some other thing involving "r" and "a", so we could just as well use that other thing (that's what it means for two things to be equal!). Hence, "Cont r a" is just a shorthand for "(a -> r) -> r".

So what is "(a -> r) -> r"? In general, for any types "x" and "y", the type "x -> y" is the type of functions which take an "x" as input and return a "y".

So "a -> r" is a function which takes an "a" and returns an "r".

So "(a -> r) -> r" is a function which takes a function from "a" to "r", and returns an "r".

riwsky• 9 years ago

In case people were wondering, the haskell stdlib answer to his interview question is at https://hackage.haskell.org/package/transformers-0.5.2.0/doc... (ContT is a generalized form which lets people run, for example, computations that involve IO). Though explaining it is still the fun part.

tempodox• 9 years ago

... prove that this type forms a monad.

How do I “prove” a type forms a monad? I only have to implement `lift` and `bind` with the correct behavior, and I have a monadic interface. Did I prove then that the type forms a monad?

MarisaKirisame• 9 years ago

Assuming "correct behavior" mean no bottom, and follow monad law, yes.

gumby• 9 years ago

While I agree with this point:

Good developers understand that they can't do everything, and they know how to leverage tools as prosthetics for their brains.

it can also become a distracting fetish.

The Soft Side of Software

Related Stories