Learning Languages Followup; Test Languages

I managed to write a new language, and 135 unit tests in that language, in a single day. When I talked about big wins coming from writing in the correct language, this is what I meant.

So, the language is a test language for a function library. We have arithmentic, date, string, and other functions that we need to test. Each function is identified internally by a GUID, and may be configured. So the language looks like;

01 #
02 # Check Array Sum function
03 #
04 declare Sum = {11669A5A-45BA-46c0-A6F6-97CDE4F5CAA5}
05 Sum(null) = null
06 Sum([]) = 0
07 Sum([1.0, 2.0]) = 3.0

In this short script, I add comments, give a function a name (binding the name `Sum` to the function identified with the id `{11669A5A-45BA-46c0-A6F6-97CDE4F5CAA5}`. Then, I define three tests; `Sum(null) = null` means what you would expect; call the sum function, passing in a single null parameter; the result should be null.

Having defined this language (which, I think, took me about an hour) I was then able to write about 135 tests with relative ease. The equivalent C# unit tests would be full of repetition and would not express their meaning anywhere near as fully. You’ve have something like;

[TestMethod]
public void TestSumNullIsNull()
{
var expected = (double)null;
var thefunction = FieldModifierHost.Instance()[“{11669A5A-45BA-46c0-A6F6-97CDE4F5CAA5}”];
var maker = thefunction.MakeMethod();
var instance = maker(new object[]{});
var result = instance(null);
Assert.AreEqual(expected, result);
}

Which is frankly impenetrable.

PS: I’ve just had a colleague add a number of tests, without any instruction, and he’s managed to put confidence tests around a function he wants to change in minutes. Unit test languages FTW!

‘This’ in Javascript and C#

I noticed something today while learning jQuery, and that’s the way the keyword `this` differs between C# and JavaScript. It suprised me when I saw some javascript that looked like;

01 $(document).ready(function() {
02 $(‘div’).each(function() {
03 this.style.color = ‘blue’;
04 });
05 ));

and I realised that this wouldn’t work in C# — at least, not the same way it works in JavaScript. In the JavaScript above, the `this` on like 03 refers to each `div` element that’s being iterated over.

Now consider similar C# code;

class Document
{
List divList = …;

void Ready()
{
divs.foreach(delegate () {
this.style.color = “blue”;
});
}
}

In C#, `this` doesn’t refer to the div, but to the Document class.

In both pieces of code, we’re creating a function with a reference to `this`, but they mean different things;

– In C#, `this` means `the object that declares the function`
– In JS, `this` means `the object the function is being invoked on.`

To see the difference, realize that you can attach the same function to two different javascript objects, and you’ll see `this` referring to each one in turn. Here’s a piece of javascript to illustrate;

var func = function() {
alert(this.name);
}

var obj1 = { ‘name’: ‘first object’, ‘func’: func };
var obj2 = { ‘name’: ‘second object’, ‘func’: func };

obj1.func();
obj2.func();

When you run this; you get two alerts: `first object` and `second object`. But when you run this in C#

Action func = delegate() {
MessageBox.Show(this.GetHashCode());
};

var obj1 = new { func = func };
var obj2 = new { func = func };

obj1.func();
obj2.func();

You see the same hashcode in both message boxes. It’s the hashcode of the object that contains this method.

So. Don’t confuse the meaning of `this` in C# and JavaScript. They are very different beasts.

Now, if you want C#’s semantics in Javascript, you have to take account of this behaviour. With my C# head on, I was tempted to understand ‘`this`’ as a _variable name_, but it isn’t. It’s a keyword, and not a variable name. To make it work like C#, you need to create a _real_ variable, and use the variable in the function. Like so;

var outerThis = this; // declare a real variable
func = function() { alert(outerThis.name); }

And this will give you C# semantics in Javascript.

Python-style string formatting for C#

[Jon Skeet][skeet] recently asked in one of [his posts][op]:

> it would be really nice to be able to write:
>
> `throw new IOException(“Expected to read {0} bytes but only {1} were available”, requiredSize, bytesRead);`

Which would do the same as

throw new IOException(String.Format(
“Expected to read {0} bytes but only {1} were available”,
requiredSize, bytesRead));

And it got me wondering about the String.Format method, and how much uglier it makes C# code to read than, say, the equivalent python code. Alongside each other;

// C#
string message = String.Format(
“Expected to read {0} bytes but only {1} were available”,
requiredSize, bytesRead);

// python
message = “Expected to read %s bytes but only %s were available” % (requiredSize, bytesRead)

I think I’d solve the problem, not by creating a new constructor for `IOException`, but by making String.Format part of the C# syntax. It works very nicely for python, and it’s such a common thing to do that I tink it would warrant a change to the language. Given how cumbersome String.Format is, it’s often shorter and clearer to use simple string concatenation. This makes things rather inconsistent.

Here’s what I came up with. It’s a ‘first draft’, and more for interest’s sake than as something I’d put into production.

Instead of passing an object array in as the values, I’m reading from the properties of an object. So you can do it with objects or tuples;

var person = new Person()
{
firstname=”Steve”,
secondname=”Cooper”
};

Then you can inject the tuple into a format string like this;

string message = “{firstname} {surname} says injecting properties is fun!”.ㄍ(person)
// message == “Steve Cooper says injecting properties is fun!”

So you’ll see this weird thing on the end of the format string that looks like a double-chevron. This is supposed to look like a double arrow, pushing values into the format string. In fact, it’s the [Bopomopho letter ‘G’][g] and therefore a perfectly normal C# method name.

Here’s the code for the double-chevron method. I say again, this is _just a proof of concept_, not production code. Use at your own peril. (In fact, don’t use. Write your own. It’ll be more solid.)

public static class StringFormatting
{
public static string ㄍ(this string format, object o)
{
var rx = new System.Text.RegularExpressions.Regex(@”{(?w+)}”);
var match = rx.Match(format);
while (match.Success)
{
string name = match.Groups[“name”].Value;
format = format
.Replace(“{“, “{{“)
.Replace(“}”, “}}”)
;
format = format.Replace(“{{” + name + “}}”, “{0}”);

object prop = o.GetType().GetProperty(name).GetValue(o, null);
format = string.Format(format, prop);
match = rx.Match(format);

}
return format;
}
}

[skeet]: http://msmvps.com/blogs/jon_skeet/default.aspx
[op]: http://msmvps.com/blogs/jon_skeet/archive/2009/01/23/quick-rant-why-isn-t-there-an-exception-string-params-object-constructor.aspx
[g]: http://www.alanwood.net/unicode/bopomofo.html

SonicFileFinder for Visual Studio

For those of you who use Visual Studio all day, can I suggest that you install [SonicFileFinder][sff]?

[sff]: http://jens-schaller.de/blog/2008/12/15/295.htm

This lovely little addin by Jens Schaller gives you a way to find files in your current solution with a few keypresses. Invoke it, and you see a dialogue like this;

Sonic File Finder

Type in a fragment of a filename, and you’ll get a filtered list of files matching that fragment. Choose a file, hit ‘return’, and the file opens in the code editor.

Basically, if you know the name of your file, you no longer need to use the Solution Explorer. As codebases get bigger and bigger, this addin gets more valuable as the Solution Explorer gets worse.

Highly recommended, plus it now works with F#, C#, and VB.NET projects.

Modifying large codebases in dynamic and static languages

I’ve been wondering recently about dynamic languages, and static languages, and the relative benefits.

I’m struggling with this question because I write C#3 by day, and am learning python in the evenings. I’m only writing small python scripts at the moment and I’d like to write larger pieces, but I’m concerned about how easy it’ll be to make certain types of change.

For example. You’ve got 100,000 lines of code. You also have a logging function that’s looks like this;

void Log(string message)

And it’s called about 200 times in your code. You decide you need a severity; so you change the signature to

void Log(string message, LoggingSeverity severity) { .. }

Now, how long does it take to find all the calls to the Log() function that need to be updated? Under C#, about ten seconds. Once every call has been fixed, the code is almost certain to work correctly.

Consider, on the other hand, the python function

def log(message):

What happens if you change the signature to

def log(message, severity):

There is no way to tell where the log message is called. You’ve just introduced 200 bugs.

It’s made even worse by duck typing; maybe you have two loggers — a deployment logger which writes to a database, and a test logger which writes to stdout. You update the database logger so it has severity. Your tests continue to pass, but your deployed system will fail.

So it seems to me that static languages give you much more power to make changes to large codebases. I’d love to know if, and where, the mistakes are in my thinking.

Object-oriented vs class-oriented programming

In his well-reasoned blog post, [chuck hoffman argues][ch] that what are normally called object-oriented programming languages should probably more rightly be called class-oriented languages. The distinction hopefully becomes clear when you consider this example.

[ch]: http://nothinghappens.net/?p=214

You are modelling people, and you want to create a person type. You should be able to strike up a conversation, so we want a ‘greet’ method for each. Our people (Alice, Bert, Charlie, and Dennis) all respond differently;

– Alice responds to a greeting with “Hi!”, or a surly “what!?” if she hasn’t had her morning coffee.
– Bert responds with either “don’t bother me, I’m walking Spot” or “what can I do for you?”, depending on whether he is walking his dog.
– Charlie responds with either “good morning”, “good afternoon”, or “good evening”, depending on the time of day.
– Dennis responds with “Hello, world!”

Now, in C#, that’s really tricky. Each person uses a different function to answer your greeting. But in C#, the Person class can only have one implementation. You could munge them all together;

class Person
{
string Greet()
{
if (isBert && isWalkingSpot) {return “don’t bother me, I’m walking Spot”; }
else if (isAlice && !hasHadCoffee) { return “what!?”; }
… etc
}
}

But that is monstrous. You could create subclasses;

class Dennis: Person
{
public override string Greet() { return “Hello, World!”; }
}

But this isn’t a class of thing; Dennis is singular. There’s not a whole class of Dennises, just a single solitary one.

What you really want to be able to do is something like this; (excuse the made-up syntax)

Person Alice = new Person();
Alice.HasCoffee = false;
Alice.Greet = { (HasCoffee ? “Hi!” : “what!?”) }

Person Dennis = new Person();
Dennis.Greet = { “Hello, World” };

That’s what an object-oriented, rather than a class-oriented, version of c# might look like.

C# Coding; Missing Functions on IEnumerable

Me old mucker Spencer pointed out today that C# 3’s newly-refurbished IEnumerable<T> class lacks some basic features. Specifically, it lacks equivalents for the classic Map, Filter, and Reduce functions seen in functional languages. The first two are familiar as List<T>.ConvertAll, and List<T>.FindAll. The third isn’t so familiar, but is still very useful. I’ve also thrown in an implementation of ForEach for free.

[Ben Hall](http://blog.benhall.me.uk/2007/08/converting-ienumerable-to-ienumerable.html) points out that it’s possible to extend the class, but I wanted to get a full, commented implementation of the three functions. Feel free to use this code in your own work.

So, here they are;

IEnumerableExtras
——

public static class IEnumerableExtras
{
///

/// Do ‘action’ to every item in the list.
///

/// The source type
/// the IEnum
/// the action to perform.
public static void ForEach
(this IEnumerable list, Action action)
{
foreach (T item in list) { action(item); }
}

///

/// Convert every item in the list using the converter
/// function
///

/// The source type
/// The destination type
/// the list to convert
/// a function to convert
/// one item to another.
/// all items in the list converted by
/// the converter function.
public static IEnumerable Map
(this IEnumerable list, Converter converter)
{
foreach (T item in list)
{
yield return converter(item);
}
}

///

/// Returns a new enumerator containing only those
/// elements which return true from ‘condition’.
///

/// The source type
/// the list to filter
/// the ‘keep in’ condition
/// the items for which condition(item)
/// is true
public static IEnumerable Filter
(this IEnumerable list, Predicate condition)
{
foreach (T item in list)
{
if (condition(item))
{
yield return item;
}
}
}

///

/// Reduces a list of items to a single item; can be
/// used to, say, sum a list of integers, or
/// concatenate a number of strings, or find the
/// maximum value in a collection.
///

///
///
///
///
public static T Reduce
(this IEnumerable list, Func reducer)
{
IEnumerator enumerator = list.GetEnumerator();
if (enumerator.MoveNext())
{
// we have some items; start combining them together.
T aggregator = enumerator.Current;
while (enumerator.MoveNext())
{
aggregator = reducer(aggregator,
enumerator.Current);
}
return aggregator;
}
else
{
// there was nothing in the list; return default.
return default(T);
}
}
}

And here’s an **example program**;

static void Main(string[] args)
{
IEnumerable maybeDoubles =
new List {1, 2, null, 3, 4, null, null, null};

// remove all the empty values: [1,2,3,4]
IEnumerable noNulls = maybeDoubles.Filter(x => x.HasValue);

// convert Nullable to non-nullable: >[1,2,3,4]
IEnumerable notNullable = noNulls.Map(x => x.Value);

// convert to strings so we can display them. [“1”, “2”, “3”, “4”]
IEnumerable stringVersions = notNullable.Map(x=>x.ToString());

// join the strings together with commas “1, 2, 3, 4”
string displayString = stringVersions.Reduce( (s1, s2) => s1 + “, ” + s2);

// show us the result;
Console.WriteLine(displayString);
Console.ReadLine();
}

So there you go. Enjoy.