The official Fatica Labs Blog! RSS 2.0
# Wednesday, 25 April 2012

Yesterday I had an interesting discussion with a colleague about using IoC in the real world in a proper way, without falling back in the Service Locator Antipattern, and of course without having a reference to the container, that is just so horrible to not worth any word. So we both found the solution being the use of factories, supplied by the container, to create the on-the-fly requested components. I was used to craft these factories by hand, but I know is a little silly, especially when the colleague talked about the Castle(Windsor) typed factories. So quite interesting, but since I choose NInject as my IoC ( Castle was the first container I saw, and my first love too ) I felt the challenge to implement the same but… Fortunately ( well, for say it would be nice being the implementer ) Remo Gloor provided a NInject extension that does almost the same for this other beautiful container. A nice documentation is also available on the Github repository. So thanks to a discussion I improved my programming knowledge, the Ninject.extensios.factoy is added to my toolbox A bocca aperta

Wednesday, 25 April 2012 15:58:04 (GMT Daylight Time, UTC+01:00)  #    Comments [0] - Trackback
CSharp | Programming

# Wednesday, 29 February 2012

In this post I will show how to make testable something that ( at least me ) usually left as untested. I’m talking about the preparing phase of a console app, the checking arguments error reporting and so on. That logic is usually so simple that any good cow boy programmer would probably leave outside any unit testing. Unfortunately we should at least do some manually check that prove that logic working, and doing things manually is always silly. In this post we assume we have a working command line parsing library, and a mocking framework. Let see the cowboy code:

        static void Main(string[] args)
        {
            string optA,optB;
            optA = optB = null;
            bool done = false;
            OptionSet set = new OptionSet();
            set.Add("a=", (k) => optA = k);
            set.Add("b=", (k) => optB = k);
            set.Add("h", (k) => { LongHelp(); done = true; });
            set.Parse(args);
            if (done)
                return;
            if (string.IsNullOrEmpty(optA) || string.IsNullOrEmpty(optB))
            {
                ShortHelp();
                return;
            }
            DoTheJob(optA,optB);
            
        }

        private static void DoTheJob(string optA, string optB)
        {
            //something interesting here
        }

        private static void LongHelp()
        {
            Console.Error.WriteLine("Long help here...");
        }

        private static void ShortHelp()
        {
            Console.Error.WriteLine("Short help here");
        }
    }

So nothing special, the example is actually very simple, we have two mandatory parameters, a command line switch to print a long help. If one argument is missing a short help line must be presented. If all the parameters are provided, the DoTheJob() method should be called with the correct values.

Current code is not testable without hosting the console application as a process, and looking at the stdout to see what happen. Even by this strategy, we can not punctually check what is passed to DoTheJob. So we want to refactor the code, without adding any complexity to the app. So here below the proposed refactoring:

    public class Program
    {
        static void Main(string[] args)
        {
            new Program().Run(args);
        }
        public virtual void Run(string[] args)
        {
            string optA, optB;
            optA = optB = null;
            bool done = false;
            OptionSet set = new OptionSet();
            set.Add("a=", (k) => optA = k);
            set.Add("b=", (k) => optB = k);
            set.Add("h", (k) => { LongHelp(); done = true; });
            set.Parse(args);
            if (done)
                return;
            if (string.IsNullOrEmpty(optA) || string.IsNullOrEmpty(optB))
            {
                ShortHelp();
                return;
            }
            DoTheJob(optA, optB);

        }

        public virtual void DoTheJob(string optA, string optB)
        {
            //something interesting here
        }

        public virtual void LongHelp()
        {
            Console.Error.WriteLine("Long help here...");
        }

        public virtual void ShortHelp()
        {
            Console.Error.WriteLine("Short help here");
        }
    }

 

So pretty easy, we provide a non static method Run(), and all the internal function are declared virtual. This is a five minutes modification we could probably apply to any other code like this we have. The difference is that we can write some unit test, lets see how:

        [TestMethod]
        public void ShouldDisplayShortHelp()
        {
            var moq = new Mock();
            moq.CallBase = true;
            moq.Setup(k=>k.DoTheJob(It.IsAny(),It.IsAny()))
                .Throws(new InvalidProgramException("Should not call"));
            moq.Object.Run(new string[0]);
            moq.Verify(k => k.ShortHelp());
        }
        [TestMethod]
        public void ShouldDisplayLongHelp()
        {
            var moq = new Mock();
            moq.CallBase = true;
            moq.Setup(k => k.DoTheJob(It.IsAny(), It.IsAny()))
                .Throws(new InvalidProgramException("Should not call"));
            moq.Object.Run(new string[]{"-h"});
            moq.Verify(k => k.LongHelp());
        }
        [TestMethod]
        public void ShouldInvokeWithProperParameters()
        {
            var moq = new Mock();
            moq.CallBase = true;
            moq.Setup(k => k.DoTheJob("p1", "p2")).Verifiable();
            moq.Object.Run(new string[] { "-a=p1","-b=p2" });
            moq.Verify();
        }

 

I used the MoQ library, please note the Callbase set to true, because we are using the same object for driving and for expect calls. So in conclusion, we achieve a real unit test of something we sometimes left apart, we did that in memory, and even if the example is really trivial, the concept can be used in complex scenarios too. What about testing the inside part of DoTheJob()? well, if a good testing strategy is used, the internal part should be testable outside somewhere else, here we are  proving we can test the shell. 

Wednesday, 29 February 2012 21:47:36 (GMT Standard Time, UTC+00:00)  #    Comments [0] - Trackback
CodeProject | CSharp | Programmin

# Saturday, 17 December 2011

I would like to present here a little argument verification library that does not require you to type any string for specifying the name of the parameter you are checking. This lets the library faster to use, not intrusive in the actual method code, and refactor friendly. As a bonus you can use it by just embedding a single file. We can see below an example, just to get immediately to the point:

As we can see, there is no magic string at all. All the argument name are guessed thanks to the metadata contained in the linq Expression we use. For example the method at line 14 if called with a null value will report:

Value cannot be null.
Parameter name: arg1

The same happens to the more complex check we do at line 46, when we write:

Contract.Expect(() => array).Meet(a => a.Length > 0 && a.First() == 0);

We have a complex predicate do meet, described by a lambda, standing that the input array should have first element zero, and non zero length. Notice that the name of the parameter is array, but we need to use another name for the argument of the lambda ( in this case I used ‘a’ ), the library is smart enough to understand that ‘a’ actually refers to array, and the error message will report it correctly if the condition does not meet. Just to clarify, the message in case of failure would be:

Precondition not verified:((array.First() == 0) AndAlso (ArrayLength(array) > 1))
Parameter name: array

Well it is not supposed to be a message to an end real user, it is a programmer friendly message, but such validation error are supposed to be reported to a developer ( an end user should not see method validation errors at all, should he ? )

Well Meet is a cutting edge function we can use for complex validations. Out of the box, for simpler cases we have some functions too, as we can see on the IContract interface definition:

An interesting portion of the codebase proposed is the one renaming the parameter on the lambda expression, to achieve the reported message reflect the correct offending parameter. It is not so easy because plain string replacement would not work:we can have a parameter named ‘a’, seen in any place in the expression string representation and a plain replacement would resolve in a big mess, furthermore Expressions are immutable. So I found help on StackOverflow, and a reply to this question solved the problem, let see the “Renamer” at work ( Thanks to Phil ):

Basically is a reusable class that take the new name of the parameter and returns a copy of the input expression with the (single) argument changed.

To improve the library or just use it, please follow/check out the project on Bitbucket, suggestions and comments are always welcome.

Saturday, 17 December 2011 13:24:25 (GMT Standard Time, UTC+00:00)  #    Comments [0] - Trackback
CodeProject | CSharp | Linq | Recipes

# Friday, 09 December 2011

It is easy to interface the Google data API by using the library Google  supply, for .NET too. Let’s have an example of acceding the contacts information:

 

This class requires the following references:

  • Google.GData.Client
  • Google.GData.Contacts
  • Google.GData.Extensions

All these are available in precompiled form after installing the Google Data API setup. Of course the complete API contains method to interact with a loot of good things in addition:

  • Blogger
  • Calendar
  • Calendar Resource
  • Code Search
  • Contacts
  • Content API for Shopping
  • Documents List
  • Email Audit
  • Email Settings
  • Google Analytics
  • Google Apps Provisioning
  • Google Health
  • Google Webmaster Tools
  • Notebook
  • Picasa Web Albums
  • Spreadsheets
  • YouTube

The only missing point: there is not (yet) a version for WP7, and the current codebase is not easy to port. Another missing point is that the API does not support OAuth2, that is indeed supported by the Google platform itself.

Friday, 09 December 2011 12:34:07 (GMT Standard Time, UTC+00:00)  #    Comments [0] - Trackback
CodeProject | CSharp | google-api

# Saturday, 03 December 2011

Since I had an amazing number of views on my previous article about my chess engine rewriting and publishing it OS, I decided to extend a little bit more the discussion. Unfortunately this is not a brand new argument, since there is a lot of good articles on the web, but in order to me some missing point exists: if you start reading the code of a fully fledged engine, even in C#, you will probably get lost in a big mesh of heuristics and optimizations without really get what’s really happens. By contrary, if you read the literature you will find a lot of pseudo code but nothing really working, and something that is a detail for the pseudo code, can be really difficult to implement in real life just to see what’s happens. Here we will show how a plain algorithm from the literature behave in it’s essence, solving a real chess problem. Of course this will not works in a real playable engine but it has a big advantage: it is *understandable* and can be the starting point to optimize, so by gradually reaching the fully fledged engine we eventually get each single steps better.

Which algorithm use ? Chess engines uses some flavor of an algorithm called MiniMax, with an immediately ( even for a simply case ) necessary optimization called Alpha Beta Pruning. This is what we will show by example here below. So what exactly is MiniMax ? It is an algorithm that works by producing a tree of the possible games in which each node is a possible status and each arc that produce the transaction is the move ( the decision ) the player can do. At each node we can weight the result of the player Mini and the player Max, Mini win if that value is little, and Max win when the value is high, so Mini want to *minimize* a score function, and Max want to maximize it. Since chess is a symmetric game, we can say that a good result for Mini is a bad result for Max and vice-versa. This lead us to a single evaluating function, with sign changed depending on the player. This simplification is referred in literature as Negamax.  Lets see an example of a game tree, by starting from a specific chess position (2rr3k/pp3pp1/1nnqbN1p/3pN3/2pP4/2P3Q1/PPB4P/R4RK1 w - - 0 0):

     image

The position is our root node, and a portion of the resulting tree is:

image

Well it is a portion, its impossible to draw it all even for just a few play, it is even impossible computationally enumerate all nodes eve for a few ply, because of the high branching factor chess has. The branching factor is a measure on how many nodes are generated from a root, in other word, in chess is an average count of the possible moves a board has. For chess this number is about 35, and so we have, for each ply an exponentially increasing number of nodes like 35^n, where n is the number of ply. Let’s consider too why it is so important having a correct move generator: just a single wrong move somewhere will mess an enormous amount of nodes.

 

 

 

 

 

average number of nodes per ply in chess:

1 35
2 1225
3 42875
4 1500625
5 52521875
6 1838265625

Of course this is just average data, can be even worst in some situation. You can always know the exact count of nodes by using the perft test contained in the same project, but I suggest you to start with a 5/6 ply and see how long it takes befor tryng 8/9 ;)

So some optimization is necessary since such an exponential explosion can’t be managed with any kind of CPU. The only game I know in which generating all the tree is probably tic-tac-toe, but for chess is absolutely not the case. So we introduce alpha beta pruning in our algorithm, but how can we prune some nodes despites to other? let’s have an example with the same position shown above, and suppose we move the Knight in c6 ( Nxc6), the black can catch it with the rock, or with the pawn, Rxc6 and  bxc6 respectively. In an alpha beta pruning scenario as soon such a move refute the white move, ie the move give a gain better than the current opponent better score, the search stops at that level. This is an enormous gain in term of performance, the only draw back is that we have just a lower bound of the actual score of a position, so we don’t really know if we can do better, but we stay on the fact that we can do enough. How this is achieved by code? Let see what we need:

  1. A way of score the position: material balance is more than enough for this sample.
  2. An algo that traverse the algo keeping track of the best score for a player ( alpha ) and for the opponent ( beta )
  3. A way to sort the move ordered so the “strongest” are seen first, the weak later.

Point 1 is easy, just give some value to each piece type, and sum it considering positive the white if the white is the player or vice-versa. The algorithm we will see soon, but the tricky part is the 3). As you probably guess, having good move navigated first, increment the changes of stops the search ( the so called beta-cut off ) with a dramatic performance increment. So the first real heuristic that will give your engine strength and personality is that function. In the example we will use a very basic ordering strategy, that put all promotion and good capture in front, all the “other” moves in the center, and the bad captures at the end. ( a good capture is one in which the catcher has less value or equal to the captured ).

So let’s show the “Vanilla” algorithm. Why “vanilla” ? because a real chess engine extends a lot this concepts,and add lot of other functionality to make the engine responsive, but the one shown do the job and it is ( hopefully ) as clear as understand as the pseudo code, whit the difference that it is working code you can inspect and debug and use for learn:

The interesting portion are the Search function. I used delegates to extract the non algorithm related code so it appear simple as pseudo code, but it is working. Then I wrote a test case using this search function here:

 

[TestMethod]
       public void TestQg6()
       {
           using (var rc = new RunClock())
           {
               var engine = new SynchronEngineAdapter(new SimpleVanillaEngine(7),
                   "2rr3k/pp3pp1/1nnqbN1p/3pN3/2pP4/2P3Q1/PPB4P/R4RK1 w - - 1 1"
                   );
               Assert.AreEqual("g3g6", engine.Search());
               Console.WriteLine("Elapsed milliseconds:" 
                   + rc.GetElapsedMilliseconds());
           }
       }

 

 

The code of the search is called by the class SimpleVanillaEngine, this is just a wrapper that inject the proper move generation  calls and evaluation/ordering functions. That test works in about 40 sec on my laptop, that is unacceptable for a real engine, but satisfying because… even if the code is simple, it report the correct answer, why can I say so ? because the board I proposed is some sort of standard test  for chess engines. Please note that the correct move Qg6 is reported in the test as g3g6 since our engine does not yet supports the human algebraic notation, but the move as you can guess is equivalent. This case is important because it show how an apparently wrong move can lead in a win if we look deep enough.

Well if interest in the project continue as it started, I will blog again on how to move this in a real engine.

Saturday, 03 December 2011 13:03:44 (GMT Standard Time, UTC+00:00)  #    Comments [0] - Trackback
Chess | CodeProject | CSharp | Games

# Saturday, 26 November 2011

I decide to publish my chess engine plays on bitbucket. Well it is almost a redo from scratch of a complete but buggy chess engine I wrote in the past that I decided to rewrite just because it was difficult to stack into the old code what I learned. The version present when I write this post contains just the move generator and the complete test for it ( I used an other nice engine: roce to compare my perft test results against. What is a perft test ? Well it is a test to prove our engine produces, from a starting board, all the possible different boards in a certain number of ply, accordingly to the chess game rules. This test also give an idea on how fast is the strategy we use to generate moves, even if this can affect just in part the overall performance of the alpha/beta pruning, we should not write a slow blobby monster. Lets see below a session of the test working:

image

The strange string showing the position are board situations expressed in FEN Notation, that is almost the standard notation we use to talk about board situations. How many test does FelpoII move generator passes ? Well here is the file containing the FEN boards, with the depth plies move counts shown, there is a lot of positions, even tricky and generally challenging positions ( in term of rules ).

What is the performance ? Since almost all chess engine are written in C++ or in C, can a C# engine works at the same level of magnitude of performance ? Here below the performance we have for the starting position:

Depth: 6 119060324 moves 5,34 seconds. 22283422,048 Move/s. Hash Hit=1400809

The same test with roce ( that is a C++ chess engine ):

Perft (6): 119060324, Time: 4.208 s

So almost the same, that is good if we remember that we wrote in C# Smile We just used a little hack: as you probably know ( or will know if you will start playing with chess engines development ) chess engines uses hash tables to store information about a board ( by using hashes against Zobrist Keys ) this tabled is sored in an unamanaged big memory array, this achieved a really sensible increase in performances.

Well some more details about the engine:

  • It uses a 0x88 board representation
  • It uses object oriented code ( so it is easier to understand compared to traditional C++ engines )
  • The internal random numbers for the zobrist key are generated by a Mersenne Twister Generator, that really solved some nasty bug due to wrong hash conflicts when I used the standard random algo of c#.
  • It uses a trasposition table in unmanaged memory to increase performance.
  • It has a performance comparable ( at least in move generation ) with traditional C/C++ engines

 

What we can do next ?

Complete the engine with a good working Negamax alpha – beta pruning algorithm.

What can we do with the code as is ?

We can use the move generator as is to validate game moves in a two human player UI, or generating fancy images from FEN positions, write a WPF ( another ) chess board ( winboard compatible? so a lot of engines are already written for it) and so on.

Enjoy.

Saturday, 26 November 2011 10:11:13 (GMT Standard Time, UTC+00:00)  #    Comments [3] - Trackback
CodeProject | CSharp | Games

My Stack Overflow
Contacts

Send mail to the author(s) E-mail

Tags
profile for Felice Pollano at Stack Overflow, Q&A for professional and enthusiast programmers
About the author/Disclaimer

Disclaimer
The opinions expressed herein are my own personal opinions and do not represent my employer's view in any way.

© Copyright 2017
Felice Pollano
Sign In
Statistics
Total Posts: 157
This Year: 0
This Month: 0
This Week: 0
Comments: 124
This blog visits
All Content © 2017, Felice Pollano
DasBlog theme 'Business' created by Christoph De Baene (delarou) and modified by Felice Pollano