Anton Gochev's Weblog

Think twice before to implement some idea – # 1

ganton — Thu, 09 Jul 2009 17:24:35 +0000

I’ve already written some posts that show some source code that can be rewritten in a different way and this way to achieve better performance. Now I will open a series with a topic of this post with a number and I’ll post different samples of implementations that I’ll run into and I’ll do some simple tests in order to find which is better. I’m not an expert of the performance improvement and for sure not all my conclusions are correct but let me share my opinion.

Sometime ago I had to improve the performance of a part of the source code in one of our applications and during this process I decided to cache in-memory some data frequently used during the operation that I was optimizing. I needed all the collected information in small arrays by parent id in a hash table. Initially, I used a list to prepare the array and after that each time to call List.ToArray() in order to use the information (it is not important why it is a matter of other implementations in the application). Then I’ve decided to prepare the arrays at the beginning and to save time in their future use. I wrote a code similar to those below.

Dictionary<int, Item[]> dictionary = new Dictionary<int, Item[]>();
for (int i = 0; i < items.Length; i++)
{
    Item item = items[i];
    Item[] currentItems;
    if (dictionary.TryGetValue(item.ParentId, out currentItems))
    {
        Item[] newCurrentItems = new Item[currentItems.Length + 1];
        Array.Copy(currentItems, newCurrentItems, currentItems.Length);
        newCurrentItems[newCurrentItems.Length - 1] = item;
        dictionary[item.ParentId] = newCurrentItems;
    }
    else
    {
        dictionary.Add(item.ParentId, new Item[] { item });
    }
}

Yesterday, a colleague of mine Saso Jordanoski ask me why I didn’t use a List in order to prepare the arrays. We’ve started a conversation and I’ve explained why but the point of Saso was that I re-allocate so much memory creating a lot of unusual arrays and that it will be better to use a List. Yes, in this case we think it is true but during the discussion we came up with an idea to just use the array of items with sorted by parent id array of items and to copy from it. And do you know? The result of my simple test shows that the last idea is much better, almost 8 times :). Think twice before to implement an idea! Below is the implementation of the last idea.

int count = 0;
for (int i = 0; i < items.Length; i += count)
{
    Item item = items[i];
    count = Count(items, i, item.ParentId);
    Item[] currentItems = new Item[count];
  
    Array.Copy(items, i, currentItems, 0, count);
    dictionary.Add(item.ParentId, currentItems);
}

It even looks simpler and easy to be understood. Below is the fool code of the test application.

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Diagnostics;
  
namespace CopyToTest
{
    class Program
    {
        static void Main(string[] args)
        {
            Item[] items = new Item[100000];
  
            for (int i = 0; i < 20000; i++)
            {
                int startIndex = i * 5;
                for (int j = startIndex; j < startIndex + 5; j++)
                {
                    items[j] = new Item { ParentId = i, Value = string.Empty };    
                }                
            }
  
            Stopwatch sw = new Stopwatch();
            sw.Start();
            Dictionary<int, Item[]> dictionary = new Dictionary<int, Item[]>();
            for (int i = 0; i < items.Length; i++)
            {
                Item item = items[i];
                Item[] currentItems;
                if (dictionary.TryGetValue(item.ParentId, out currentItems))
                {
                    Item[] newCurrentItems = new Item[currentItems.Length + 1];
                    Array.Copy(currentItems, newCurrentItems, currentItems.Length);
                    newCurrentItems[newCurrentItems.Length - 1] = item;
                    dictionary[item.ParentId] = newCurrentItems;
                }
                else
                {
                    dictionary.Add(item.ParentId, new Item[] { item });
                }
            }
            sw.Stop();
  
            Console.WriteLine(sw.ElapsedMilliseconds);
  
            sw.Reset();
            sw.Start();
            dictionary = new Dictionary<int, Item[]>();
            int count = 0;
            for (int i = 0; i < items.Length; i += count)
            {
                Item item = items[i];
                count = Count(items, i, item.ParentId);
                Item[] currentItems = new Item[count];
  
                Array.Copy(items, i, currentItems, 0, count);
                dictionary.Add(item.ParentId, currentItems);
            }
            sw.Stop();
  
            Console.WriteLine(sw.ElapsedMilliseconds);
            Console.Read();
        }
  
        private static int Count(Item[] array, int startindex, int parentId)
        {
            int count = 0;
            for (int i = startindex; i < array.Length; i++)
            {
                if (array[i].ParentId == parentId)
                {
                    count++;
                }
                else
                {
                    break;
                }
            }
            return count;
        }
    }
  
    public class Item
    {
        public int ParentId;
        public string Value;
    }
}

And you can see the results on my machine below.

is operator + object casting or as operator + null comparison?

ganton — Sat, 27 Jun 2009 20:44:46 +0000

This week I found in the source code of one of our projects that people use “is” operator + object casting in places that they can use “as” operator + null comparison.

// "is" operator + object casting
if (myObj is MyObject)
{
    MyObject obj = (MyObject)myObj;
}
  
// "as" operator + null comparison
MyObject obj = myObj as MyObject;
if (obj != null)
{
    // do some work with obj
}

What is the problem? The problem is that in the first case you cast the object two times which is slower than if you cast the object only one time as in the second case. Why it happens? Because the “is” operator casts the object behind the scene and if the cast is successful it will return true, otherwise false. Consequently, the first case is slower than the second case. You may suggest that nowadays the hardware is very fast and you will be correct but the problem is that most of the developers that use the first case even don’t understand why is better to use the second one. Below is the test sample for the performance that I did. The count is a very high number that will allow us to see the difference.

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Diagnostics;
  
namespace AsIsTest
{
    class Program
    {
  
        public class MyObject
        {
            public MyObject Parent { get; set; }
        }
  
        static void Main(string[] args)
        {
            int count = 100000000;
            object myObj = new MyObject { Parent = new MyObject() };
            Stopwatch watch = new Stopwatch();
            watch.Start();
            for (int i = 0; i < count; i++)
            {
                MyObject obj;
                if (myObj is MyObject)
                {
                    obj = (MyObject)myObj;
                }
            }
            watch.Stop();
            Console.WriteLine("is and cast time if true: " + watch.ElapsedMilliseconds);
  
            watch.Reset();
            watch.Start();
            for (int i = 0; i < count; i++)
            {
                MyObject obj = myObj as MyObject;
                if (obj != null)
                {
  
                }
            }
            watch.Stop();
            Console.WriteLine("as and null comparison time: " + watch.ElapsedMilliseconds);
  
            Console.Read();
  
        }
    }
}

The result is below.

Note that in the sample above both operators are able to cast the project. In this case there is a performance advantage of the second sample but if the object cannot be casted you will get the similar results for both cases because only one casting takes place in the first sample. Below is a sample when the object can not be casted.

watch.Start();
for (int i = 0; i < count; i++)
{
    string obj;
    if (myObj is string)
    {
        obj = (string)myObj;
    }
}
watch.Stop();
Console.WriteLine("is and cast time if true: " + watch.ElapsedMilliseconds);
  
watch.Reset();
watch.Start();
for (int i = 0; i < count; i++)
{
    string obj = myObj as string;
    if (obj != null)
    {
  
    }
}
watch.Stop();
Console.WriteLine("as and null comparison time: " + watch.ElapsedMilliseconds);

And the result is below.

It’s funny but on my machine the results are completely the same. I’ve started the application several time and they are the same.

SQL Server Setup failed to obtain system account information for the ASPNET Account….

ganton — Fri, 19 Jun 2009 14:05:18 +0000

Today, during installation of MS SQL Server 2005 DE on my new work machine i ran into an installation error that stated “SQL Server Setup failed to obtain system account information for the ASPNET Account.To proceed, reinstall the .Net Framework, and then run SQL Server Setup again.”. And it even offer me to reinstall the .Net Framework and then to retry to install the MS SQl Server.

Please do not follow this message directly if you have installed the .Net Framework. The problem in my case was that I have installed the framework on the machine but ASP.Net didn’t installed. In other words you just need to run aspnet_regiis.exe with a switch –i. Then you can run iisreset in order to be sure that all changes have taken an effect. It will install the runtime and all other things and you can continue with the installation.

Enjoy your installation of MS SQL Server 2005 DE.

Configure linked servers for Excel and MS SQL Server in MS SQL Server 2005

ganton — Wed, 22 Apr 2009 08:47:33 +0000

Linked servers allows

A linked server configuration allows Microsoft® SQL Server™ to execute commands against OLE DB data sources on different servers

Quote from MSDN

This way we have an abstract layer to other data sources (different from MS SQL Server) and we can connect them using T-SQL of MS SQL Server and only one instance of it. In addition, we can join tables from different resources which is completely transparent for our stored procedures or ad hoc queries.

In this post I’ll show you how you can configure a linked server for Excel file and a linked server for a MS SQL Server database. Assume that we have an Excel file named people.xls as those below

We also have a MS SQL Server database named test with one table named test. Well, the names are not too descriptive but for the sample are good enough.

Assume that we’d like to use the information from the excel file and also to extend it with some information like address and city of the person. I added external_id column into the table in order to be able to join the table with the Excel file. This field contains the same values as id column of the excel file.

Before to configure linked services you should ensure that DTC (Distributed Transaction Coordinator) Service is started.

You can configure both using the Microsoft SQL Server management Studio but I’ll use a query to do so. SQL Server contains a stored procedure named sp_addlinkedserver. It allows you to create linked servers.

   1: exec master.dbo.sp_addlinkedserver    2:     @server = N‘Excel’,    3:     @srvproduct=N‘Excel’,    4:     @provider=N‘Microsoft.Jet.OLEDB.4.0′,    5:     @datasrc=N‘C:\Program Files\Microsoft SQL Server\MSSQL.1\MSSQL\Data\people.xls’,    6:     @provstr=N‘Excel 5.0′;

The @server parameter is the name that we will use to connect the server. The @srvproduct is not required here but left it with a value of Excel. The @provider parameter shows which OLE DB provider we gonna use. In our case it is Microsoft.Jet.OLEDB.4.0 provider. The @datasrc is the full path to the Excel file and @provstr should be Excel 5.0. After executing the procedure we can access the linked server. I’ll use sp_tables_ex stored procedure to retrieve all sheets (tables) that current linked server contains.

   1: exec sp_tables_ex @table_server=N‘Excel’

The result is below.

Note that you should not use the Excel file with any other program when you are using it from the SQL Server. If you do so you will get an exception like those below.

Msg 7399, Level 16, State 1, Procedure sp_tables_ex, Line 41

The OLE DB provider "Microsoft.Jet.OLEDB.4.0" for linked server "Excel" reported an error. The provider did not give any information about the error.

Msg 7303, Level 16, State 1, Procedure sp_tables_ex, Line 41

Cannot initialize the data source object of OLE DB provider "Microsoft.Jet.OLEDB.4.0" for linked server "Excel".

Having the linked server named Excel we can query the file using a script like those below.

   1: select * from Excel…Sheet1$

The result is below.

Now let’s connect our SQL Server database as a linked server. To do so I’ll use the same stored procedure as before but with a different OLE DB provider.

   1: exec master.dbo.sp_addlinkedserver    2:     @server = N‘SQL’,    3:     @srvproduct=N”,    4:     @provider=N‘SQLNCLI’,    5:     @datasrc=N‘GOCHEVA\SQLEXPRESS’;

We are still not ready to connect to the SQL linked server. Before trying to access it we should allow remote connections to the target SQL Server Instance. You can complete this action via SQL Server 2005 Surface Area Configuration as shown below.

Now I can connect it and get all databases that are the SQL Server instance contains and I also can get their table information. Using sp_catalogs stored procedure I’ll get all DBs.

   1: exec sp_catalogs @server_name=N‘SQL’

And the result.

I’ll use sp_tables_ex to get all tables of a specific catalog (database) of the SQL linked server.

   1: exec sp_tables_ex    2:     @table_server=N‘SQL’,    3:     @table_schema=N‘dbo’,    4:     @table_catalog=N‘test’;

And the result.

Now, we can use an as hoc query using T-SQL and to retrieve a joined information from both files.

   1: select s.*, t.address, t.city    2: from Excel…Sheet1$ as s, SQL.test.dbo.test as t   3: where s.id = t.external_id

And the result is.

That’s all for now. have fun using SQL Linked Server option.

MS Technology Specialist: WSS 3.0 - Application Development

ganton — Mon, 20 Apr 2009 08:40:29 +0000

Last Friday, I passed MCTS: WSS 3.0 - Application Development successfully. I’d like to share my experience with this exam in this post.

First of all, it was really hard to prepare myself for the exam because I’ve started to learn and practice WSS 3.0 last December. I’ve passed several Microsoft exams and this one was the most complicated one for me. I suppose that it is because I have a good theoretical knowledge but I have not enough practice with WSS 3.0.

What have I seen on the exam? First of all, I was a bit surprised when I’ve read that the exam includes several questions that are not scored and they are for test purposes of the MS exam improvement. Actually, I cannot remember about such questions in the exams that I’ve passed in the past. In any case if you are prepared good enough it will not be a problem. [Update - a friend of mine Martin Karaivanov told me that there are also such questions in other MS exams.] The second thing was a question(s) about Records Repository Web Service. I cannot say that I was surprised because I read in the Internet that such questions may appear but as you may know Records Repository Web Service is more like a part of MOSS 2007 than a part of WSS 3.0. Most of the questions of the exam were connected with WSS 3.0 Object Model, Features, Definitions, Solutions, Web Parts and how to install them, custom controls, custom list columns, custom content types, etc. You also should have some knowledge of stsadm.exe and a very good knowledge of CAML.

How did I prepare myself? I’ve used only information from the Internet including on-line training, videos and tutorials and some of the books recommended by Microsoft on their page of Preparation Guide for Exam 70-541. You can find some Internet resources in my previous posts in WSS category. A very important part of my preparation was that I was practice myself in the use of WSS 3.0 Object Model, Web Services, configuring WSS 3.0 via Object Model and Administrative Web Site and learning and practicing CAML.

Manage WSS 3.0 meetings workspaces and meetings via Web Services

ganton — Fri, 27 Mar 2009 09:22:41 +0000

WSS 3.0 platform provides a bunch of Web Services that can be used to manage different WSS objects. They expose via Web Services interfaces part of the functionality that is available for you via WSS 3.0 object model. The full list of available Web Services you can find here.

In this post I’d like to share with you my experience of using Meetings Web Service in order to create, remove, update a meeting workspace and add a new meeting to the created workspace.

Before starting you should create a new project in Visual Studio and to add a Web Reference to the Meetings.asmx WSS Web Service. For this sample we will use a WPF Application project. Below is the main form of the application that we will use in order to provide an interface of managing meeting workspaces to the end-user. It allows the end-user to specify WSS web site URL, meeting workspace title, choose a template, etc.

Now we are ready to write a code that will create a new meeting workspace. To do so, we will use CreateWorkspace method of the Web Service. This method will create a new meeting workspace subsite on the specified server running WSS. The method accepts four parameters: workspace’s title, the template name that will be used when the workspace is created, the locale identifier and an instance of TimeZoneInf class. You can go to MSDN documentation and find what templates are supported for meeting workspace and use one of them. But in our solution I choose to retrieve all available workspace templates from the Web Service and to give to the end-user an option to choose one of them. To do so, we will use GetMeetingsInformation method of the service. It accepts two parameters: request flags and locale ID. Using request flags we can control what of the information available for the workspace we’d like to retrieve. In this sample we will use an unsigned integer 4. It will require from the service to provide to us available templates information. Below is the code that we can use in order to get all available templates and put them into a hashtable and a combobox.

   1: proxy.Url = ComposeUrl();               2: XmlNode information = proxy.GetMeetingsInformation(UInt32.Parse("4"), UInt32.Parse("1033"));   3: templatesList.Clear();   4: templateTypeComboBox.Items.Clear();   5:     6: foreach (XmlNode node in information.ChildNodes[0].ChildNodes)   7: {   8:     templateTypeComboBox.Items.Add(node.Attributes["Title"].Value);   9:    10:     templatesList.Add(node.Attributes["Title"].Value,  11:         node.Attributes["Name"].Value);  12: }  13: if (templateTypeComboBox.Items.Count > 0)  14: {  15:     templateTypeComboBox.SelectedIndex = 0;  16: }

Assume that we have defined and initialized a Meetings Web Service instance named proxy somewhere on the class level and we can use it everywhere in the class code. ComposeUrl method will return a full Url of the meetings.asmx service depending of what WSS web site Url the end-user has chosen. The important moment here is the call of the GetMeetingsInformation method with a request flag 4. Below is the XML response that the method returns.

   1: xml version="1.0" encoding="utf-16"?>   2: <MeetingsInformation xmlns="http://schemas.microsoft.com/sharepoint/soap/meetings/">   3:   <ListTemplates>   4:     <Template   5:       Name="MPS#0"   6:       Title="Basic Meeting Workspace"   7:       Id="2"   8:       Description="A site to plan, organize, and capture the results of a meeting. It provides lists for managing the agenda, meeting attendees, and documents."   9:       ImageUrl="http://gocheva-vm1/_layouts/images/mwsprev.png" />  10:     <Template  11:       Name="MPS#1"  12:       Title="Blank Meeting Workspace"  13:       Id="2"  14:       Description="A blank meeting site for you to customize based on your requirements."  15:       ImageUrl="http://gocheva-vm1/_layouts/images/blankmwsprev.png" />  16:     <Template  17:       Name="MPS#2"  18:       Title="Decision Meeting Workspace"  19:       Id="2"  20:       Description="A site for meetings that track status or make decisions. It provides lists for creating tasks, storing documents, and recording decisions."  21:       ImageUrl="http://gocheva-vm1/_layouts/images/decisionmwsprev.png" />  22:     <Template  23:       Name="MPS#3"  24:       Title="Social Meeting Workspace"  25:       Id="2"  26:       Description="A site to plan social occasions. It provides lists for tracking attendees, providing directions, and storing pictures of the event."  27:       ImageUrl="http://gocheva-vm1/_layouts/images/socialmwsprev.png" />  28:     <Template  29:       Name="MPS#4"  30:       Title="Multipage Meeting Workspace"  31:       Id="2"  32:       Description="A site to plan, organize, and capture the results of a meeting. It provides lists for managing the agenda and meeting attendees in addition to two blank pages for you to customize based on your requirements."  33:       ImageUrl="http://gocheva-vm1/_layouts/images/multipagemwsprev.png" />  34:   ListTemplates>  35: MeetingsInformation>

Note that we have several templates returned with additional information in