I initially used ISO 10126 padding, but that seemed to produce a different cipher in C# and Java. I decided to go with no padding which leads me to the  cryptographic exception: Length of the data to decrypt is invalid. After some head banging I settled with PKCS7. 

Same encoding of strings has to be used in java and c#, other wise you will end up having different ciphers in java and c#.

The java code below uses a base64 util class from android SDK but you can replace it like with one from apache commons

C# encryption utility

public RijndaelManaged GetRijndaelManaged(String secretKey)
        {
            var keyBytes = new byte[16];
            var secretKeyBytes = Encoding.UTF8.GetBytes(secretKey);
            Array.Copy(secretKeyBytes, keyBytes, Math.Min(keyBytes.Length, secretKeyBytes.Length));
            return new RijndaelManaged
            {
                Mode = CipherMode.CBC,
                Padding = PaddingMode.PKCS7,
                KeySize = 128,
                BlockSize = 128,
                Key = keyBytes,
                IV = keyBytes
            };
        }

        public byte[] Encrypt(byte[] plainBytes, RijndaelManaged rijndaelManaged)
        {
            return rijndaelManaged.CreateEncryptor()
                .TransformFinalBlock(plainBytes, 0, plainBytes.Length);
        }

        public byte[] Decrypt(byte[] encryptedData, RijndaelManaged rijndaelManaged)
        {
            return rijndaelManaged.CreateDecryptor()
                .TransformFinalBlock(encryptedData, 0, encryptedData.Length);
        }

        /// <summary>
        /// Encrypts plaintext using AES 128bit key and a Chain Block Cipher and returns a base64 encoded string
        /// </summary>
        /// <param name="plainText">Plain text to encrypt</param>
        /// <param name="key">Secret key</param>
        /// <returns>Base64 encoded string</returns>
        public String Encrypt(String plainText, String key)
        {
            var plainBytes = Encoding.UTF8.GetBytes(plainText);
            return Convert.ToBase64String(Encrypt(plainBytes, GetRijndaelManaged(key)));
        }

        /// <summary>
        /// Decrypts a base64 encoded string using the given key (AES 128bit key and a Chain Block Cipher)
        /// </summary>
        /// <param name="encryptedText">Base64 Encoded String</param>
        /// <param name="key">Secret Key</param>
        /// <returns>Decrypted String</returns>
        public String Decrypt(String encryptedText, String key)
        {
            var encryptedBytes = Convert.FromBase64String(encryptedText);
            return Encoding.UTF8.GetString(Decrypt(encryptedBytes, GetRijndaelManaged(key)));
        }

Java Encryption Utility

private final String characterEncoding = "UTF-8";
	private final String cipherTransformation = "AES/CBC/PKCS5Padding";
	private final String aesEncryptionAlgorithm = "AES";

	public  byte[] decrypt(byte[] cipherText, byte[] key, byte [] initialVector) throws NoSuchAlgorithmException, NoSuchPaddingException, InvalidKeyException, InvalidAlgorithmParameterException, IllegalBlockSizeException, BadPaddingException
	{
		Cipher cipher = Cipher.getInstance(cipherTransformation);
		SecretKeySpec secretKeySpecy = new SecretKeySpec(key, aesEncryptionAlgorithm);
		IvParameterSpec ivParameterSpec = new IvParameterSpec(initialVector);
		cipher.init(Cipher.DECRYPT_MODE, secretKeySpecy, ivParameterSpec);
		cipherText = cipher.doFinal(cipherText);
		return cipherText;
	}

	public byte[] encrypt(byte[] plainText, byte[] key, byte [] initialVector) throws NoSuchAlgorithmException, NoSuchPaddingException, InvalidKeyException, InvalidAlgorithmParameterException, IllegalBlockSizeException, BadPaddingException
	{
		Cipher cipher = Cipher.getInstance(cipherTransformation);
		SecretKeySpec secretKeySpec = new SecretKeySpec(key, aesEncryptionAlgorithm);
		IvParameterSpec ivParameterSpec = new IvParameterSpec(initialVector);
		cipher.init(Cipher.ENCRYPT_MODE, secretKeySpec, ivParameterSpec);
		plainText = cipher.doFinal(plainText);
		return plainText;
	}

	private byte[] getKeyBytes(String key) throws UnsupportedEncodingException{
		byte[] keyBytes= new byte[16];
		byte[] parameterKeyBytes= key.getBytes(characterEncoding);
		System.arraycopy(parameterKeyBytes, 0, keyBytes, 0, Math.min(parameterKeyBytes.length, keyBytes.length));
		return keyBytes;
	}

	/// <summary>
	/// Encrypts plaintext using AES 128bit key and a Chain Block Cipher and returns a base64 encoded string
	/// </summary>
	/// <param name="plainText">Plain text to encrypt</param>
	/// <param name="key">Secret key</param>
	/// <returns>Base64 encoded string</returns>
	public String encrypt(String plainText, String key) throws UnsupportedEncodingException, InvalidKeyException, NoSuchAlgorithmException, NoSuchPaddingException, InvalidAlgorithmParameterException, IllegalBlockSizeException, BadPaddingException{
		byte[] plainTextbytes = plainText.getBytes(characterEncoding);
		byte[] keyBytes = getKeyBytes(key);
		return Base64.encodeToString(encrypt(plainTextbytes,keyBytes, keyBytes), Base64.DEFAULT);
	}

	/// <summary>
	/// Decrypts a base64 encoded string using the given key (AES 128bit key and a Chain Block Cipher)
	/// </summary>
	/// <param name="encryptedText">Base64 Encoded String</param>
	/// <param name="key">Secret Key</param>
	/// <returns>Decrypted String</returns>
	public String decrypt(String encryptedText, String key) throws KeyException, GeneralSecurityException, GeneralSecurityException, InvalidAlgorithmParameterException, IllegalBlockSizeException, BadPaddingException, IOException{
		byte[] cipheredBytes = Base64.decode(encryptedText, Base64.DEFAULT);
		byte[] keyBytes = getKeyBytes(key);
		return new String(decrypt(cipheredBytes, keyBytes, keyBytes), characterEncoding);
	}

Hopefully this saves someone time.

I was hesitant to learn android development because of time. Last week, a requirement kicked in that needed an android front-end. Target platform was android 2.2.  This is my feel of android development after six days of using android SDK. Actually I had also never used an android phone, so I was amazed how cool android OS  is.I should say mobile development should have been easier like how android development is from the beginning. I have developed in QT, and J2ME, but both dont match up to android development with ease. The great UI controls that come out of the box, the eclipse integrated plugin, the ability to reuse most of the standard java code, its all great.

I did consult the SDK documentation and the hello tutorials during the development. They where of great help. The application pulls the data from a .NET WCF JSON Rest based service. I used the Google Gson for the serialization and de-serialization of json to object types.  On the WCF service I decorated by operation contracts with WebGet and WebInvoke depending on whether am making GET or POST requests. To reuse the forms based authentication, the http connection service takes care of  managing the cookies mainly the forms based authentication cookie to receive and send it on requests.

Coding in android though seams easier than QT or j2me, I got some beating for not reading the documentation. Some of the beating I got:-

• Embedding more than one root view in the scroll viewer was just throwing exception. I was quite big headed on googling for a solution on this one, because I just couldn’t expect it was the cause.
• The table layout. This I had to abandon and choose another approach. I needed to replicate a telerik like silverlight gridview with a details view on each row. Was t0o ambitous about it, but abandoned it. Got headaches about the table row layout parameters and scrollview.
• Adding an activity to the manifest file. Was always forgetting about this step and that led me to debugging. And pretty much I would remember (f***) when the exception is raised from the line where I start the Activity.
• A point I got so frustrated with exceptions from setting content view. The exceptions just could’t make sense. I don’t know what I f.*up, but when I change layouts, or add new layouts. I tend to get exceptions when setting activity view. It just couldn’t make sense why. so I went to the bin folder and did the mighty SHIFT-DELETE-RETURN. Damn. all references to resources in R.java couldn’t be recognized any more. I was quick to think may be its a build order, so I moved gen above src folder, build project and back. Some magic happened, no more errors even when setting the content view. I still don’t know why for now. But this seems to be the solution I do for now when I get exceptions when setting content view. It seems to work. Till when I become a sniper at android, I will overlook this.

Damn! Hooking my configuration service to android preferences activity has been the easiest way of managing preferences on a mobile app have seen.

Lastly for now. The way resources are managed in android development, I can’t find an execuse for the laziness of harding cording strings in code files. Its damn fast, easy. Android is great. Android development platform is very great. Now I just feel cheated because I was missing a lot for a long time. I must boostrap my N900 with Nitdroid.

Android development is the way mobile development should have been like from the beginning.

def calculate_distance(fingerprint1, fingerprint2):
    '''  http://oxygene.sk/lukas/2011/01/how-does-chromaprint-work/
    This algorithm is extracted from the postgres chromaprint c function
    Instead of returning the score
    I return 1-score as the distance measure
    '''
    if len(numpy.shape(fingerprint1)) != 1 != len(numpy.shape(fingerprint2)):
        raise Exception('Only one dimension arrays allowed')
    if fingerprint1 == None or fingerprint2 == None:
        return 0

    numcounts = len(fingerprint1) + len(fingerprint2) + 1;
    counts = numpy.zeros(numcounts);
    for i in range(len(fingerprint1)):
        jbegin = max(0, i - ACOUSTID_MAX_ALIGN_OFFSET);
        jend = min(len(fingerprint2), i + ACOUSTID_MAX_ALIGN_OFFSET);
        for j in range(jbegin,jend):
            biterror = popcount_lookup8(fingerprint1[i] ^ fingerprint2[j]);
            # ereport(DEBUG5, (errmsg("comparing %d and %d with error %d", i, j, biterror)));
            if (biterror <= ACOUSTID_MAX_BIT_ERROR):
                offset = i - j + len(fingerprint2)
                counts[offset] += 1

    return 1.0 - (numpy.max(counts) /(1.0* min(len(fingerprint1), len(fingerprint2))))

1. Unlock the modem using Simbwa Phillip steps.
2. Set the APN of the SIM Network whenever initializing the modem

Simbwa Phillip has some steps which I was able to follow to perfectly unlock my Huawei 3G modem.
I will just paste here the steps again so that you don’t have to bounce back and forth.

Unlock the modem using Simbwa Phillip steps

#!/usr/bin/env python

import hashlib
import humod
UNLOCK_SALT = "5e8dd316726b0335"

def show_code(imei, salt):
    digest = hashlib.md5((imei+salt).lower()).digest()
    code = 0
    for i in range(0,4):
        code += (ord(digest[i])^ord(digest[4+i])^ord(digest[8+i])^ord(digest[12+i])) << (3-i)*8
        code &= 0x1ffffff
        code |= 0x2000000
    return code

mod = humod.Modem()
imei = mod.show_imei()
status = humod.at_commands.Command(mod,'AT^CARDLOCK="'+str(show_code(imei,UNLOCK_SALT))+'"\r')
#------------------- code ends here -----------------------------

 #!/usr/bin/env python

import hashlib
UNLOCK_SALT = "5e8dd316726b0335"

def show_codes(imei, salt):
    digest = hashlib.md5((imei+salt).lower()).digest()
    code = 0
    for i in range(0,4):
        code += (ord(digest[i])^ord(digest[4+i])^ord(digest[8+i])^ord(digest[12+i])) << (3-i)*8
        code &= 0x1ffffff
        code |= 0x2000000
    return code

# PLEASE TYPE YOUR IMEI  BELOW
imei = ' '
if(imei == ' '):
    print 'Please open this script from your editor and enter your IMEI'
else:
    print "Your Modem unlock code: %s" % show_codes(imei,UNLOCK_SALT);

Set the APN of the SIM Network whenever initializing the modem

For linux if your using wvdial change the wvdial.conf or add the line below which sets the APN name

Init3 = AT+CGDCONT=1,"IP","TelecomNetworkAPNName"

For windows users, go to the device manager -> Modems -> Open properties of the modem -> Under Advanced -> in the extra initialization commands field put the entry

AT+CGDCONT=1,"IP","TelecomNetworkAPNName"

The TelecomNetworkAPNName varies from Telecom, for UTL it is utweb and for Orange it is orange.ug.
So now you can create a connection with a dial number *99# or *99***1#.

Am blogging about this because when you look at the graph of CPI items, it seems they are just increasing not going down. I look/analyze the inflation rate in Uganda using Composite Consumer Price Index (CPI) values from the Bank of Uganda Website. From wiki inflation rate is a measure of inflation, the rate of increase of a price index (for example, a consumer price index). If we can predict CPI values that are used to calculate inflation rate, that means we are able to predict the inflation rate.

The major factors (CPI items) that decide the Uganda inflation rate as stated in the excel sheet are:-

• Food
• Beverages and tobacco
• Clothing and footwear
• Rent, Fuel and utilities
• House hold and personal goods
• Transport and communication.
• Education
• Health ,entertainment & Others

If we are able to build a linear function that can predict each of the items above based on their historical values, then we can predict the inflation rate of Uganda with an acceptable Root mean square error. For each of the CPI items above, Least squares regression will be used to find five weights that we apply to recent historical values to estimate/predicate next month CPI of the item and the Root mean square error (RMSE) from predictions.

THE RED LINE SHOWS THE ORIGINAL VALUES, THE BLUE LINE SHOWS THE PREDICTED VALUE

To be able to predict the next CPI value for an Item, we take four previous values from current date and apply weights for them. e.g if we want to predict the Food CPI for march we take values Feb 2011 to Nov 2010.

2010	Nov	173.44
	Dec	173.69

2011	Jan	179.53
	Feb	183.04

The predicated Food CPI for march based on weights w5=0.022637182175 w4=0.100118564555 w3=-0.512433115279 w2=1.4003471128 w1=-0.887631618337 would be:-

183.04 * w5 + 179.53 * w4 + 173.69 * w3 + 173.44 * w2 + w1 = 175.10

After calculating the predicted CPI for every Item, them we can apply the inflation rate.
Predicting values may not give the right answer since we are only looking at values of CPI, to further bring it closer to the present value, surrounding factors e.g. currency rate have to be considered.

To be Continued…
END OF PART 1

Brute force, Generate and Test are easy to implement. but what about using Arc Consistency. For adventure I attempted to try out using arc consistency to solve sudoku puzzles. I did manage to implement the concept though my implementation is not as fast as Norvig’s.

To ease the implementation, I modeled the sudoku as follows:-

• A Cell knows about  the row, column, block, general sudoku reference and its domain.
• Both Column, Block and Row  know about the sudoku reference and the cells that are in it
• Sudoku class itself knows its blocks, rows, columns and cells.
• I also keep a list of constraints cell-to-row, cell-to-block and cell-to-column

With the arrangement above I perform arc consistency on the constraints.

import copy
import time

class Cell(object):
    row=None
    column=None
    block=None
    domain=None
    value=None
    sudoku=None
    def __init__(self,sudoku, value):
        self.sudoku=sudoku
        self.value=value
        self.domain=[]
    def __hash__(self):
        return int(str(self.row.index)+str(self.column.index))

    def __eq__(self,other):
        try:
            return self.row.index == other.row.index and self.column.index == other.column.index
        except:
            return False
    def __str__(self):
        return "Cell " + str(self.row.index) + str(self.column.index)

class Row(object):
    sudoku=None
    index=None
    cells=None
    def __init__(self,sudoku, index):
        self.sudoku=sudoku
        self.index=index
        self.cells=[]
    def addCell(self, cell):
        self.cells.append(cell)
        cell.row=self
    def __hash__(self):
        return int("9"+str(self.index))
    def __eq__(self, other):
        return isinstance(other, Row) and self.index == other.index
    def __str__(self):
        return "Row " + str(self.index)

class Column(object):
    sudoku=None
    index=None
    cells=None
    def __init__(self, sudoku,index):
        self.sudoku=sudoku
        self.index=index
        self.cells=[]
    def addCell(self,cell):
        self.cells.append(cell)
        cell.column=self
    def __hash__(self):
        return int("10"+str(self.index))
    def __eq__(self, other):
        return isinstance(other, Column) and self.index == other.index
    def __str__(self):
        return "Column " + str(self.index)

class Block(object):
    sudoku=None
    index=None
    cells=None
    def __init__(self,sudoku, index):
        self.sudoku=None
        self.index=index
        self.cells=[]
    def addCell(self, cell):
        self.cells.append(cell)
        cell.block=self
    def __hash__(self):
        return int("11"+str(self.index))
    def __eq__(self, other):
        return isinstance(other, Block) and self.index == other.index
    def __str__(self):
        return "Block " + str(self.index)

class Sudoku(object):
    columns=None
    rows=None
    blocks=None
    cells=None
    TDA = None
    __width=9

    def getBlockIndex(self, rowIndex, columnIndex):
        if rowIndex <= 2:
            if columnIndex <= 2:
                return 0
            elif columnIndex <= 5:
                return 1
            elif columnIndex <= 8:
                return 2
        elif rowIndex <= 5:
            if columnIndex <= 2:
                return 3
            elif columnIndex <= 5:
                return 4
            elif columnIndex <= 8:
                return 5
        elif rowIndex <= 8:
            if columnIndex <= 2:
                return 6
            elif columnIndex <= 5:
                return 7
            elif columnIndex <= 8:
                return 8
        return None

    def __init__(self, puzzle):
        self.TDA=dict()
        self.cells=[]
        self.blocks=[ Block(self,n) for n in range(self.__width)]
        self.columns=[ Column(self,n) for n in range(self.__width)]
        self.rows = [ Row(self,n) for n in range(self.__width)]

        for row in range(self.__width):
            for column in range(self.__width):
                cell = Cell(self,puzzle[row][column])
                self.cells.append(cell)
                self.columns[column].addCell(cell)
                self.rows[row].addCell(cell)
                self.blocks[self.getBlockIndex(row,column)].addCell(cell)

                if cell.value:
                    cell.domain.append(cell.value)
                else:
                    cell.domain.extend(range(1, self.__width+1))
                    self.TDA[(cell, cell.row)]=1
                    self.TDA[(cell, cell.column)]=1
                    self.TDA[(cell, cell.block)]=1

    def ArcConsistency(self):
        while (1 in self.TDA.values()):
            for cell, constraint in self.TDA:
                if self.TDA[(cell, constraint)] == 0: continue
                cellDomainModified=False
                for constraintCell in constraint.cells:
                    if cell != constraintCell and constraintCell.value != None and constraintCell.value in cell.domain:
                        cell.domain.remove(constraintCell.value)
                        cellDomainModified=True
                self.TDA[(cell, constraint)]=0
                if cellDomainModified:
                    for primeConstraint in [cell.row, cell.column, cell.block]:
                        if primeConstraint != constraint:
                            for primeConstraintCell in primeConstraint.cells:
                                if  primeConstraintCell.value==None:
                                    self.TDA[(primeConstraintCell, primeConstraint)]=1
        cellConstraintsToRemove=[]
        for cell, constraint in self.TDA:
            if len(cell.domain)==1:
                cell.value = cell.domain[0]
                cellConstraintsToRemove.append((cell, constraint))
        for cellConstraintToRemove in cellConstraintsToRemove:
            del self.TDA[cellConstraintToRemove]

    def solve(self):
	start=time.clock()
        self.ArcConsistency()
	if self.isSolved():
	    return self, time.clock()-start
	sol=self.search()
	return sol, time.clock()-start

    def isInconsistent(self):
	invalid=False
	for cell in self.cells:
	    if cell.value==None and len(cell.domain) == 0:
		invalid=True
		break
	if not invalid:
	    for row  in self.rows:
		if not self.areCellsValid(row.cells):
		    invalid=True
		    break
	if not invalid:
	    for block  in self.blocks:
		if not self.areCellsValid(block.cells):
		    invalid=True
		    break
	if not invalid:
	    for column  in self.columns:
		if not self.areCellsValid(column.cells):
		    invalid=True
		    break

	return invalid

    def areCellsValid(self, cells):
	noneNullCells = [cell for cell in cells if cell.value != None]
	return len(noneNullCells) == len(set(cell.value for cell in noneNullCells))

    def search(self):
	try:
	    if threading.currentThread().stopFlag:
		return None
	except:
	    pass

	sudokuCopy = copy.deepcopy(self)
	count, cell = min((len(cell.domain), cell) for cell in sudokuCopy.cells if cell.value == None)
	if( cell == None): return
	cellIndex = sudokuCopy.cells.index(cell)
	for value in copy.deepcopy(cell.domain):
	    sudokuTemp = copy.deepcopy(sudokuCopy)
	    sudokuTemp.cells[cellIndex].value=value
	    sudokuTemp.cells[cellIndex].domain=[]
	    sudokuTemp.cells[cellIndex].domain.append(value)
	    del sudokuTemp.TDA[(sudokuTemp.cells[cellIndex], sudokuTemp.cells[cellIndex].row)]
	    del sudokuTemp.TDA[(sudokuTemp.cells[cellIndex], sudokuTemp.cells[cellIndex].column)]
	    del sudokuTemp.TDA[(sudokuTemp.cells[cellIndex], sudokuTemp.cells[cellIndex].block)]
	    for tdaDictKey in sudokuTemp.TDA:
		sudokuTemp.TDA[tdaDictKey]=1
	    sudokuTemp.ArcConsistency()
	    if sudokuTemp.isInconsistent():
		continue
	    elif sudokuTemp.isSolved():
		return sudokuTemp
	    else:
		solution=sudokuTemp.search()
		if solution and solution.isSolved():
		    return solution
	return None

    def isSolved(self):
        value=True
        for cell in self.cells:
            if cell.value == None:
                value=False
                break
        return value

    def toPuzzle(self):
        return [ [ cell.value for cell in row.cells ] for row in self.rows ]

if __name__ == "__main__":

    def printSudoku(sudokuPuzzle):
	for row in sudokuPuzzle.toPuzzle():
	    print str(row)

    puzzle=[[None,None,None,None,None,None,None,None,1],
            [None,None,None,None,None,None,None,None,None],
            [None,None,None,None,None,None,None,None,None],
            [None,None,None,None,None,None,None,None,None],
            [None,None,None,None,1,None,None,None,None],
            [None,None,None,None,None,None,None,None,None],
            [None,None,None,None,None,None,None,None,None],
            [None,None,None,None,None,None,None,None,None],
            [1,None,None,None,None,None,None,None,None]]

    sudoku = Sudoku(puzzle)
    solvedSudoku, elapsed=sudoku.solve()

    index=0
    for row in solvedSudoku.toPuzzle():
	print str(row) + "\t\t\t" + str(puzzle[index])
	index=index+1
    print "Took " + str(elapsed)

So why am I happy about KwikTok? Well the 100MB of data isn’t a joke. 100MB!!!! for only 250 shillings. If you calculate that for 31 days, that is if you spend 100MB per day that totals to 7750 Shillings. That means you buying 3.1GB for just 7,750 Shillings. Hmmmm Yeaaaaaa. Now that’s sweet. 100MB is enough for you to stay online the whole day and make a couple of skype audio calls. If your to compare it with the major player Orange that am subscribed to for 3G internet 3GB would cost you 85,000 Shillings.

You should be just too crazy to underlook the UTL free internet. This means if your like me and have an old product like Orange 3G internet. You move UTL to higher priority and use Orange Internet when UTL connection is off or extremely slow. That makes you reserve your orange data when you extremely need blazing speed. Not that the UTL internet connection is slow. You can do pretty alot with the UTL connection. Pages load pretty fast and if your crazy like me who is running a proxy on all my machines and my N900 phone then that means better performance. Skype calls using my phone-n900 and laptop are both clear.

Every good thing has some annoying bit. KwikTok data doesn’t go without a negative spot. One thing that pisses me off is when your connected through the phone or your working from the phone and the 100MB is exceeded you don’t get any warning and it automatically switches to using the normal profile of your airtime which is maddly expensive. Coz my phone squashes 4000 in just less than 10minutes when active. You should be aware of this.

To be able to connect to MongoDB process you will have to download the latest C# Driver

To be able to log via the Enterprise Logging Application Block I created a custom trace listener and trace listener configuration. The trace listener configuration was to allow specifying MongoDB connection string and database name. The custom trace listener opens the connection to the MongoDB process and inserts the updates. Every attempted change to the entity is stored.

Changes to entity are logged at the point the data context is saving changes by subscribing to the SavingChanges event.


void OnSavingChanges(object sender, EventArgs e)
{
ObjectStateManager.GetObjectStateEntries(EntityState.Added | EntityState.Deleted | EntityState.Modified)
.Where(entity => !entity.IsRelationship && entity.Entity is IEntity)
.Select(entity => new AuditEntry((IEntity)entity.Entity, UserService.GetCurrentUser().Name, entity.State))
.Log();
}


The Entriprise lib LogEntry object is sent directly to the database unless it has AuditEntities. AuditEntities are saved in the AuditEntry collection whereas the rest of the log entries are stored in the LogEntry collection. The entity is saved as an attribute of AuditEntry called Entity.


public interface IAuditEntry: IEntity where TEntity : IEntity
{
TEntity Entity { get; set; }
string EntityModifiedBy { get; set; }
EntityState @EntityState { get; set; }
}


The Extension file has the code that checks the LogEntry object if it has AuditEntities in the ExtendedProperties and method for logging using the enterprise lib Logger. In the attached file I mocked (Moq) the user IIdentity object, but depending on your application context you know how to get the current user IIdentity. You can download a zip having all the files here.

• Voters Id
• Surname
• Other names
• Village
• Polling station
• Sub-Country
• Parish
• Constituency
• District
• or a combination of the above listed.

The search results reflect accurate information from the EC website http://www.ec.or.ug/ as of Febuary 19th, 2011.

If you want further analysis of the voters list:-

You can request analysis reports which I can generate for you. Depending on the complex level of the report and resources that will be required. I may take a day or two to get back to you with that report.

For advanced users like developers I allow you to write readonly App Engine mapreduce jobs that I can run on the data set. All code is first analyzed before job submitted. If interested you can contact me for the model design and further information.

NB. By the time of writing this article App Engine mapreduce doesn’t have the same feature set as Hadoop e.g no Reduce, but it provides a feature set thats pretty enough to run simple MapReduce jobs. By the way if you have cron jobs that fit being run as mapreduce jobs, I would advice you to convert to using the appengine mapreduce api because its faster and cheaper with resource quotas and time. Am not sure if it may apply to your situation, but it does to mine. I scrapped the voters list using 8 shards for more than twenty five thousand polling stations, downloading pdfs, decrypting them and inserting voter data (> 500) into the datastore. Took two days, yet the first time using cron jobs, it took almost a month.

Url: http://reg.kutokea.com
Currently the search indexes are rebuilding for more than ten million voters and will be done by the morning of 20th Febuary 2011.

• Delegates: I had to create abstract classes Action, Func, IDelegate to account for the corresponding delegates in .Net
• RunAsync: In C# it uses the BeginInvoke and EndInvoke for Java I have used the ExecutorService with a thread pool of 20.
• TimeSpan: I created a simple TimeSpan class to be used in the timing calls.
• C# Extensions: Since java doesn’t have extensions, I decided to merge the calls in AspectExtensions to the same class as AspectF

Methods I skipped converting include MustBeNonDefault and CacheList

I did use the same tests in AspectFTest class to make sure the conversion was atleast in a working condition.
For mocking I used the mockito which is as simple as Moq. The tests confirm to JUnit with some rough edge changes from xUnit.net.

The conversion wasn’t smooth since I had to hack my way around the Action and Func C# delegates.
As of now there is a hacked Alpha version which needs further improvement. Am currently using it in my small projects and will improve it further as need arises.


Download AspectF in Java
NB. NetBeans 6.9 was used to build solution