Patterns - Open / Close Principle

The first time I heard this principle is in DNRTV epis0de 71 Jean-Paul Boodhoo on Demystifying Design Patterns Part 4 .

This principle states:
An object should be Open for Extensions but Closed for Modifications.

Take a query scenario: The aim is to allow querying classes to query an object without have to change source code on the query class. In other words not having to add new methods for each new query.

This can be done by the Specification Pattern (out of box Dot Net Partterm) in particular the Predicate Delegate Pattern. (definition: Ask an Object if another Object meets a certain criteria)

Using the sample from DNRTV episode.

--------------------------------- Querying Class --------------------------------------

public void Should_find_all_employees_with_a_last_name_beginning_with_B()
{
IEmployeeContactBook book = CreateContactBook();
AddDefaultListOfEmployeesTo(book);
Assert.AreEqual(2,GetCountOf(book.AllContactsMatching(LastNameStartsWithB)));
}

public LastNameStartsWithB(IEmployee employee)
{
return employee.LastName.StartsWith("B");

}

----------------- Query Class ------------------------------------------------------

public IEnumerable AllContactsMatching(Predicate match)
{
foreach (IEmployee contact in contacts)
{
if (match(contact))
{
yield return contact;
}
}
}

Http Handlers

I have wanted to blog about HttpHandlers for a while now. At last I have found a reason, I have recently watched the DNRTV screencast 108 Miguel Castro on Http Handlers in .NET .

Http Handlers route file extensions to handlers. This is best illustrated by the fileC:\WINDOWS\Microsoft.NET\Framework\v2.0.50727\CONFIG\web.config . If you open this file you will see something like this (abbreviated due to space constraints



















So when you browse to an aspx page

the aspx extension is routed to System.Web.UI.PageHandlerFactory handler
The page class interits from the control class, which is where we get the event cycles such as onLoad and onInit etc; and also implement IHttpHandler and this has a method called ProcessRequest.

You can't browse to a master page, The reason is seen

In this case the request is routed to HttpForbiddenHandler. This bring up a form that says you cannot browse to a master page.

The extension ashx is browsable point handler. Thus it is a web page without a ui and without the web page lifecycle. The key point is that the ashx can accept query strings.

Web Control - Validators

I have watched many many DNRTV show but have never blogged about them. I'm not really sure why, as they a general very worthwhile to watch. Do I decided to go back over the archive and chose the show 75 Peter Blum on Web Controls Part 1

I chose this show as I remembered it had a lot of small interesting points. The show dealt with: -

• commonly overlooked features of ASP.NET Validators
• requiring that at least 1 textbox has text

1. commonly overlooked features of ASP.NET Validators
========================================

a) Validate a TextBox as a Date. As an aside, Personally I prefer CSLA framework concept where you can use 't' for today and 't-' for yesterday and 't+' for tomorrow etc)

Add a CompareValidor; properties;operator;change to DateTypeCheck;change type to Date

ControlToValidate="TextBox1" ErrorMessage="CompareValidator"
Operator="DataTypeCheck" Type="Date">

b) Validator Culturised (Australianised, Americanised, Ukise etc)
===============================================
Validators all follow current pages culture info object. The culture can be changed by
properties -> document -> culture

c) How to Prevent a Button from Validating
===============================
For example have a two buttons on a form (OK andCancel). You want OK to validate and Cancel not to validate.
button=>properties=>CausesValidation=>set it to false

The button will no longer provide client or server side validation.

A textbox also has a CauseValidation property. However this is part of the AutoPostBack. Thus the property name should include AutoPostBack as well as CausesValidation.

d) Server Side Validation Must be setup for it to Work
======================================
Server side validation is important to validate data against business object or database data. If you fail to do server side validation then you can be hacked. The hacked will turn off java script and then have easy access.

To setup server side validation is easy (for example):-

protected void Button1_Click(object sender, EventArgs e)
{
if (Page.IsValid)
{
//save data
}
}

2) requiring that at least 1 textbox has text
===============================

Please see the screencast for further details - as this blog is getting too long

SharePoint Content Types

After blogging recently about Sql Server, I decide today to blog about SharePoint, and in particular content types. A good resource for this is Ted Pattison screencast ContentTypes.wmv.

Content types in previous version was only possible at document library level. However, it is useful to have many different content types within a library. Thus v3 allows document types to be associated at the document level.

Step 1
=====
Before creating content types, it is important to create site columns (a stand-alone definition for site columns). Site Settings --> Galleries ->Site Columns
It is good practise to put site columns into a new group.

After creating the site columns then it is time to create the content types. The good news is that content types support inheritance. All programmer know about inheritance, but sharepoint users may not know. Inheritance allows many different content types to share properties of another content type. Thus making it easier to maintain and less initial typing.

Step 2
=====
Content types are created by Site Settings --> Galleries ->Site Content Types
Base content types is another term for the content type from which the current content type inherits. It is good practise to setup base content types if there is commonality. Thus add the base content type first.

The content type (currently the base content shall require columns). You are required to add these columns from existing columns. After finishing the create of the columns for the base content type. The derived content types should be created. This follows the same procedure as above.

Step 3
=====
The final step is to configure the document library to use these created content types. If you do not do this then under new, the only option will be to create a new folder.
In the document library go
setting -> Document Library Settings ->Advanced Settings -> click yes option fo allow management types
Note a new section content types shows up under Document Library Settings

Click add from existing site content types and select the document types for this library.
Note you may want to get rid of the default "document" document type. This is done by clicking on the "document" document type and choosing delete this content type.

That's it all Done

Sql Server Mirroring

I have just finished DAT327 - Microsoft SQL Server 2005 Always-On Demo Fest from Teched 2007. This was a marathon screencast but very worthwhile.

This was on mirroring, which I as a developer have not been responsible for. However, it was brilliant to see a comprehensive screencast on this.

The primary severs in mirroring are the: principal (primary), mirror (secondary) and witness. The witness is extremely light weight and is used for checking to see if the principal and mirror are available. The good news is the witness can run on any edition of sql server including express (FREE).

The mirror needs to be prepared as a mirroring server. On of the steps is restore the database to the mirror server and must use Recover Completion State = No Recover. No Recover says do not bring this database online, leave it inaccessible to users but able to receive transactions.

Explicit Redirection allows changing the connection string using the following syntax:
;failover partner =

The code will now try to connect to server1 or server2, and run against whichever server is available.

Say server1 is principal, server2 is mirror and server3 is the witness.

If there is a Failover, the connections are broken, transaction are rolled back, and will have to reconnect.

Scenario1: Server1 goes down - The application will work, the principal then would be server2.

Scenario2: Mirror goes down -The application will work, the principal will remain server1. However logs will grow as log file will be needed when server2 becomes available again to ensure server can catchup.

Lots of other good stuff in this screencast.

Sql Predicates

I recently found a good article on indxe tuning, but I misplaced the link. Not unusual for me. The article discussed query plans from first principles. I thus will document some of the points I recall from the article and hopefully bookmark the article next time.

For newbies I will give a definition of predicates:-
1. Grammar the part of a sentence in which something is said about the subject
2. Logic something that is asserted about the subject of a proposition

Thus in sql server it normally is part of the where cause: such as where city='Perth'. So when will an index be used to seek? The predicates cannot seek on a single column index if

• the predicate contains a function such as absolute values: where ABS([xDelta]) < 5 =" 0.">
• the predicate contains a Like expression with a wildcard in the first position
  where [city] LIKE '%erth'

In multiple column indexes the order of the keys become important. The predicate on a second column can use an index only if the first column of the index is an equality predicate. Therefore an index will be used when
where [Country]='Australia' and [City] = 'Perth'

However if the predicate is

where [Country] Like '%us%' and [City] = 'Perth' then the index will not be used.

Sql Server Data Management Views

I have recently finished watching DAT320 My Top Ten DMVs by Greg Low from Teched 2007.
The DMV are very powerful. One of the commom questions is what is the difference between DMV and a DMF. THE DMF's are DMV's with parameters.

It is important to note that additional permissions are required to run DMV. These permissions are: -

• select permission
• plus view server state (server scope)
• and view database state for (database scope)

Sql Server: Data Publishing Wizard

I recently watched DAT315 -Be More Productive with Microsoft SQL Server 2005 Tools by Paul Mestemaker from TechEd 2007. This is very difficult address topic to address as it is so vast. Thus it must be difficult to decide what to cover and in what depth. Unfortunately for me, the material addressed seemed to be jumping all over the place.

Another pet gripe of mine is lack of slides. This helps a person watching the screencast to quickly get to section in the video that are of interest. So even though 5 slides appears appealing, it really is not.

I did enjoy the reminder about the Database Publishing Wizard. I refer you to the following for more details http://blogs.msdn.com/webdevtools/archive/2007/10/15/sql-database-publishing-wizard-is-now-in-visual-studio-orcas.aspx.

The other reminder was availability of Best Practise Analyzer. This educates users on best practices by providing interactive reporting, and each rule has corresponding article.

Sql Clr - Sql Data Types

Greg Low recently presented at Sql Server and Dont Net Usergroup in Perth. He was brilliant. This encouraged me to view his webcasts from Teched 2007. The first being DAT210 where he presented on DBA guide to Clr integration.

In turn this screencast reminded me that I have not blogged about the Sql Datatypes before. Sql Clr introduced new datatypes for example SqlString, SqlBoolean etc. These datatypes are found in the System.Data.SqlTypes namespace. The difference between a SqlString and a String is that the SqlString is Nullable. That is it has a string value and also if it is null or not.

Greg suggestion is that all interfaces to Sql Server should use the Sql datatypes. In other words everytime you expose something that will be seen by Sql Server or Tsql use the Sql datatypes (inputs, types of methods, return values etc). Internal to your code does not need to do this.

Creating Unique Ids

I watched the Dat323 screencast from Teched 97 by Itzik Ben-Gan. This is definitely in my top ten of screencast. The screencast isolated 3 topics and stepped through the logic of each.

Creating unique Id's
==============
There are two types op Id's: either all the ids must always exist or missing id's are allowed. A good example of all id's must exist is invoices, while a booking number may allow for missing ids.

The method use blocking for all id's and non-blocking algorithm for missing id's.

Blocking Algorothm:-
================
CREATE TABLE dbo.SyncSeq(val INT);
INSERT INTO dbo.SyncSeq VALUES(0);
GO
CREATE PROC dbo.usp_SyncSeq
@val AS INT OUTPUT
AS
UPDATE dbo.SyncSeq
SET @val = val = val + 1;
GO

Non-Blocking Algorithm
=================
The trick here is to limit the number of rows in this table. Thus the Save allows the rollback, but still increase the increment.

CREATE TABLE dbo.AsyncSeq(val INT IDENTITY);
GO
CREATE PROC dbo.usp_AsyncSeq
@val AS INT OUTPUT
AS
BEGIN TRAN
SAVE TRAN S1;
INSERT INTO dbo.AsyncSeq DEFAULT VALUES;
SET @val = SCOPE_IDENTITY()
ROLLBACK TRAN S1;
COMMIT TRAN
GO

Perfmon Counters

I was viewing screencast DAT306 of Teched 2007 by Kevin Kline. He suggests the following counters as the most important when monitoring a Sql Server Test in Perfmon.

1. Pages/sec
2. Bytes Total /sec
3. % Disk Time
4. Avg Disk Queue Length
5. % Processor Time
6. Full Scan/sec
7. Buffer cache hit ratio
8. Transactions per sec
9. Log growths
10. User connections
11. Average latch wait time
12. Number of deadlocks per second
13. Average wait time ms
14. Memory grants pending

When you add the counter you can choose the server name to be monitored. It is highly desirable not to monitor on the same server you are testing; as the act of monitoring may skew the results.

There are many performance objects - so how do we know which to use - what is useful inour scenario? The best advise is to learn from other people, hence above Kevin Klines's suggested list.

Rss Lists

Many people think of Rss as just plain text. However RSS has the concept of list. Web323 of Teched 2007 is a well presented screencast by Eric Lawrence. The information on left is taken from that the screencast.

Many people know rss feeds because of podcasts. This form of Rss is known as enclosures. Thus enclosures include podcasts and videocasting (video). Enclosures are otherwise known as Rich Media.

The Rss Platform Api is a com api but iseasy enough to use. The classes include Feed Manager, Feed Folder, Feed, Feed Item, Feed Enclosure, Folder Events and Feed Events.

For further details refer to the screencast.

IE7 Architecture

I have been viewing a screen cast on IE7 - WEB321 from Teched 2007 by Douglas Stamper.

Particularly good on this website was the difference in architecture (with regard to privileges) see pictures and also the availability of virtual machines for IE6 and IE7.

The key aspect is the extra layers in IE7, and thus internet explorer rights have been reduced - runs with low user rights or least privileged mode. Writing to the machine or getting ino the registry is now protected. The temp internet files (tif) is where all files will be written (cached). The beauty is the tif is untrusted and trying to execute anything out of the tif will bring up a warning.

Brief explanation of w3wp.exe

I have been asked by sharepoint colleagues what is w3wp.exe. The question is usually asked when they are watching me debug.

w3wp.exe is the worker process for asp.net when using iis 6.0. In iis5.0 the worker process was called aspnet_wp.exe.

A good set of diagrams of the interelation between asp.net and iis 6.0 and iis 5.0 can be found in the screencast web304 - by Rob Howard as part of the heroes happen event.

Shared Services Best Practise

Most people know that SSP best practise is to to have only 1 instance.

However I heard it phrased well in OFC222 screen cast -

Start with one SPP and make people convince you why you will need more than 1

Mobile - Headless Device

I heard a new terminology today.

A headless device.

Never heard this before - it turns out to be a device without a display.

Nice term.

Mobile Devices - Emulator can be Accesd programatically

I have just realised that the emulator can be accessed programmatically. This is good news - the terminology that Microsoft use is "Managed Core Connectivity Wrapper" or "Managed CoreCon"

reference is Microsoft.Smartdevice.Connectivity.dll

Mobile VS2008 Security Manager

Good news to see VS2008 for Mobile has security management to facilitate testing against Mobile devices. This is required as Mobile Computers come with a lot of different security management.

Under Tools / Device Security Manager - we can set configurations that reflect the windows mobile device.

It also has certificate management.

Sharepoint SP 1 error

While installing WSS SP1 my virtual machine appeared to Hang. I eventually had toi shut down. On restarting, and trying to rerun the installation I was informed that the WSS SP1 had already been run. I the tried to install Moss SP1 but alas, I got another message informing me that there was already an installation occurring.

When I went to Sharepoint Central Admin - I got the error message regarding missing confifugarion database.

To the rescue:-

psconfig -cmd upgrade -inplace b2b -wait -force

I checked the upgrade.log in wss12 path and at the bottom found

SPFarm=, recursive = True, 0 errors and warnings encountered

Now I only wish Microsoft could easily show us that SP1 has been installed.

Sharepoint Crawl Fail /Errors

I have found on certain installation the Sharepoint crawl fails.

The primary reason I have discovered is the service
Office Sharepoint Server Search is running with local system rights. This seems to be inadequate therefore change it run under Administrator privileges.