Computers are dumb

re: Computers are dumb

Jacob — Fri, 09 May 2008 21:36:58 GMT

The first questioner should look into Take/TakeWhile. Example:

var smallNumbers = Enumerable.Range(0, int.MaxValue).TakeWhile(n => n < 10);

re: Computers are dumb

Mark Sowul — Fri, 09 May 2008 21:54:19 GMT

"Even leaving the obvious efficiency concern aside, the latter code reads better. It more clearly reflects the intention of the programmer."

Isn't that half of the point of LINQ, to allow for a limited degree of functional and declarative programming such that the "what" becomes more important than the "how"?

Unless you have a significant, measurable performance difference, write what you mean!

re: Computers are dumb

Richard — Fri, 09 May 2008 22:12:24 GMT

@Mark

"Unless you have a significant, measurable performance difference, write what you mean!"

You don't have to measure the performance to know that iterating to 2^31 in order to get 10 values is going to be inefficient. Efficiency versus readability is a continuous trade-off, can't always easily be fixed after the fact, and it's not just asymptotic differences that matter. If I can make my product 3x faster by making all my code 20% less readable (by, for instance, writing it in C++ rather than in Java), and that speed increase matters enough, then I'd better be aware of that before I write a single line of code.

Premature optimization is bad, but believing that all optimization is premature is bad too.

re: Computers are dumb

Steve — Fri, 09 May 2008 23:45:49 GMT

@Jacob: Was going to suggest to same, developers just don't know what to use most of the time :(

IEnumerable.Any() vs. IEnumerable.Count() != 0

Jerome Laban — Sun, 11 May 2008 01:24:22 GMT

An english version is available here . Après avoir lu ce post d'Eric Lippert , je me suis rappelé que

re: Computers are dumb

Dean Harding — Mon, 12 May 2008 02:20:26 GMT

"Accessing event logs from a remote machine over a slow network"

What kind of developer wouldn't have seen that coming?

re: Computers are dumb

Eric Lippert — Mon, 12 May 2008 03:41:20 GMT

A novice?

Developers have to start somewhere, you know.

re: Computers are dumb

B0rG — Mon, 12 May 2008 11:45:17 GMT

I'm wondering about this one:

if (0 != customers.Count()) ...

if (customers.Any()) ...

does Count() always count all the records in the collection, or number of items is stored somewhere?

re: Computers are dumb

James Hart — Mon, 12 May 2008 12:16:34 GMT

@B0rG - well, if customers implements IList, then the Enumerable.Count() method takes a shortcut and reads the Count property. But for a general IEnumerable, there's no guarantee that the extension Count() method will find a quicker way to count it than to enumerate the whole collection. And it's perfectly possible to implement an IEnumerable that never terminates, so Count() on a general IEnumerable may simply never return.

Calling Count() is essentially asking 'how many times do I have to call MoveNext() on an enumerator before I get an answer of 'false'?'. asking for 'Any()' is simply asking 'does the first call to MoveNext() on an enumerator return true?'

re: Computers are dumb

Tom Kirby-Green — Mon, 12 May 2008 13:05:02 GMT

The for-loop iteration is faster I belive because for built in System.Array's the JITer is able to emit efficient machine code that doesn't do range checking for each an every index look-up.

I believe for this to work the boundary check must be inline to the 'for', that is to say:

int[] data = new int[] { ... };

for (int i = 0; i < data.Length; i++)

Process(data[i]); /// Details boring

rather than:

int[] data = new int[] { ... };

int length = data.Length;

for (int i = 0; i < length; i++)

Process(data[i]); // Details boring

But like all undocumented optimizations it would dangerous to over rely on this JIT optimization.

IEnumerable.Any() vs. IEnumerable.Count() != 0

Jerome Laban — Wed, 14 May 2008 02:04:03 GMT

An english version is available here . Après avoir lu ce post d'Eric Lippert , je me suis rappelé que

VCS Team Links for May 22, 2008

Charlie Calvert's Community Blog — Thu, 22 May 2008 22:12:34 GMT

Rather than place the links to the most recent C# team content directly in Community Convergence , I