Until recently, I assumed that web applications were based on a strict client-server / request-response model. This has several limitations, such as a thin client (your web browser) not being able to do any *real* business logic. For example, say you need to validate against a back-end data source such as ensuring a username-password combo is valid or checking if a city exist within a state. As this is really outside the scope of JavaScript, it would require a hit to the backend, and then re-posting the page.
But what if you could just do a "postback" for the desired field only, leaving the rest of the page unchanged? And what if you could have the server automatically update the client without the client initiating the event?
A technology, Asynchronous JavaScript and XML (AJAX) looks like it can do just that. Quoting from Adaptive Path's Jesse James Garrett:
[Quote]
Ajax isn't a technology. It's really several technologies, each flourishing in its own right, coming together in powerful new ways. Ajax incorporates:
standards-based presentation using XHTML and CSS;
dynamic display and interaction using the Document Object Model;
data interchange and manipulation using XML and XSLT;
Previously I offered some tips for configuring Enterprise Library. Today I'll discuss another problematic area - adding strong names.
First, this builds on concepts about Strong Names and PublicKeyTokens. You may need to add a strong name to the EntLib blocks if you include it in a project that already uses strong names (because this requires that all assemblies being referenced also have strong names).
To add a strong name, you'll need to create a key and reference it in the AssemblyInfo.cs file.
Strong Names and Versioning Tips
0 - You will need to add the key to the source code. Microsoft gives us the source code partially so that we can do things like this.
1 - Put properties (like strong-naming) in the GlobalAssemblyInfo.cs. This requires you to remove (or comment out) the key from the other 38 AssemblyInfo files, however it lets you change it in one place.
2 - Note that there are two places that DLLs are stored.
Opening the EntLib solution and compiling creates copies in the \src dir. You still need to copy these to the \bin Make sure to run the "Copy Assemblies to Bin Directory" script included in the EntLib start menu. If you're not putting DLLs in the GAC, EntLib's config tool pulls DLLs from the \bin dir. So you'll notice strong-name nightmares if you don't do this. It will set the assembly name (like PublicKeyToken) from those DLLs, not merely the ones used in your project. For example, check the version of the DLLs loaded by the config tool, and make sure that they include the strong name.
Enterprise Library is Microsoft's next generation of Application Blocks. It's a free, integrated suite of common architectural components that your project probably needs. It has several good perks:
Includes all source code
An auto-config tool to handle the Xml config files.
38 projects of well written code - i.e. a good example to learn from
When I first was playing around with it, I had trouble with creating strong names and config files.
Configuration Tips
1. A single application can have multiple physical config files:
CachingQuickStart.exe.config (app)
cachingConfiguration.config (other block)
dataconfiguration.config (other block)
Furthermore, you'll need to make sure that the correct file is copied to the app's bin. Merely rebuilding or using EntLib's config tool doesn't copy all configs.
2 - For the data access block config, you'll need to set both the connection string as well as the individual params. Config looks like:
3 - For the Caching Block's Data Provider:
"PartitionName" does not correspond to actual database partitions, but rather a value in the Caching.CachData.PartitionName column. This means that you don't need to create a DB partition for the DB cache to work.
Make sure that you run the script to create the Caching DB ('Caching').
Make sure that user has access to the 'Caching' database, and can insert into the CacheData table. Merely running the create script doesn't provide access to custom users.
4 - For the Caching Block's Isolated Storage, this uses System.IO.IsolatedStorage. You do not provide a physical file path name. Partition Name is a folder name like "MyApp".
When I first came to .Net I tried to avoid this, but it actually serves a useful purpose: specifying your assemblies in the config ensures that your app is using the correct version of an assembly. Many developers I've known don't want to deal with details like that, so I'm going to blog about some things that I've found to make it easier. Where does this info come from?
This is the assembly's FullName. You could view it with the following code:
public string GetFullName(string strAssemblyPath) {
Assembly a = Assembly.LoadFrom(strAssemblyPath));
return a.FullName;
}
It has the following parts:
Assembly Name
Version
Culture
PublicKeyToken (generated from a strong-name)
This link on MSDN provides more info about Assembly names. How can I tell the version and PublicKeyToken of an assembly?
You can tell both via the Assembly.FullName property as mentioned above. But you can also use other tools.
Assembly - right click on the assembly in Windows Explorer, and look at the "Version" tab.
PublicKeyToken - use the Strong Name tool (sn.exe), run "sn -T C:\Temp\Common.dll" with the BIG 'T' (not little 't', which gives the token for the inline file instead).
My app doesn't need to worry about this. How can I simplify my life?
If you remove a section from the config file, then it won't filter on that. For example, if you don't care about the strong name, then you can simply remove it and use this instead:
However, this could get you in trouble if multiple assemblies with different strong names are used later on. Next...
I'll discuss some practical application of these concepts in a future post.
There are a lot of other great and free tools out there besides just NUnit, NAnt, and NDoc. I'm always keeping my eyes out for freeware tools because they can really simplify one's life. Two things I do to help organize this: I have a special folder in IE dedicated to just tools, and each new tool gets its own subfolder. This provides a place if I ever come upon other articles or links for that tool. If I dump every tool-related download, FAQ, and article, then the folder gets unmanageable. Once you get 20+ tools, this is a big deal. I also have a corresponding folder structure on my hard drive where I can save the installers. I don't download them directly to the ProgramFiles folder because it gets flooded.
Here are some tools that aren't as popular as NUnit, but I've still found useful.
ASP.Net includes several validator controls. These controls first validate data on the client using JavaScript. Much of the scripting for this is contained in the WebUIValidation.js file, which is in an IIS directory like \aspnet_client\system_web\1_1_4322. Each page containing any validator already includes this script, therefore you can reuse any of its functions in your own controls. For example, view the source of an aspx.page that has a validator on it, and you'll see a reference to the WebUIValidation.js file. Because it's a bunch of free methods, lets look at this file to see how we can benefit from it.
The file has many functions. I've included descriptions for some of the more relevant ones:
ValidatorUpdateDisplay(val)
ValidatorUpdateIsValid()
ValidatorHookupControlID(controlID, val)
ValidatorHookupControl(control, val)
ValidatorGetValue(id) - gets the value of a control given its Id
ValidatorGetValueRecursive(control)
Page_ClientValidate()
ValidatorCommonOnSubmit()
ValidatorEnable(val, enable) - can enable/disable a specific validator
CompareValidatorEvaluateIsValid(val) - used by Compare Validator
CustomValidatorEvaluateIsValid(val) - used by Custom Validator
RegularExpressionValidatorEvaluateIsValid(val) - used by Regular Expression Validator
ValidatorTrim(s) - trims excess white space from input
RequiredFieldValidatorEvaluateIsValid(val) - used by Required Field Validator
RangeValidatorEvaluateIsValid(val) - used by Range Validator
ValidationSummaryOnSubmit()
The ID (such as in the ValidatorGetValue) indicates a string. The 'val' parameter indicates the Validator control itself, not just the ID. I previously wrote about how to use the ValidatorEnable to Disable Validators on the Client. However you can use other methods to get the value of a control or extend existing validator functionality. This is almost essential if you're going to build your own validators by deriving from BaseValidator.
JavaScript can be messy to many developers, and it's helpful to know that such a toolbox of production-ready scripts is already available to your pages.
Like many Microsoft developers, I've used VSS as my primary source control system - and suffered for it. Perhaps the biggest problem I've had with VSS is the inability to batch the creation of ASP.Net web projects. Such applications require a virtual directory, and it seems like VSS hard-codes this info into a binary (not even an XML file) somewhere.
Automating Web project creation is useful for improving process. For example, you could create a batch script that get a version (whether the latest of a branch) from source control, and does all the preparation work to make it ready for development (set config values, create its own databases, make the virtual directories, etc...)
Normally to automate ASP.Net app creation (which I'll discuss more in another blog), you need to:
Get the source code
Create the virtual directory in IIS (with VBS, Directory Services, or even NAnt's tag).
Set the virtual directory in the *.webinfo file
Set the virtual directory in the solution file
The problem becomes when VSS is tightly integrated into Visual Studio, then it creates a fifth step:
Find that binary file, "decode" it, set the virtual directory, and recode.
After *hours* of looking at this, I finally had to move onto other things.
However, my new firm (Ameripay) uses non-VSS source control that isn't tightly integrated into VS.Net, so it doesn't have that limiting 5th step. Consequently, I can now automate the creation of Web projects.
Every now and then I find myself wanting to dynamically create an Excel Workbook in .Net. Maybe it's for a data export, or to set up a structured template. Either way, there's only a small set of common utilities needed to get started. However, because Excel's automation model is based in COM (not .Net), it can be difficult to create those utilities. So below are the ones I've started used. Two disclaimers:
I have only used these for simple, internal tools, not enterprise-critical production environments. Therefore they are not streamlined for performance.
I've collected snippets here and there from other articles (which I no longer know where), so the idea isn't original.
My goal is to be able to programmatically create an Excel sheet, like so:
ExcelHelper.Utilities u = new ExcelHelper.Utilities(); u.AddSheet("Test1"); u.SetCellContent("A",new Cell("B",3), "Hello World!"); u.SetCellContent("A",new Cell("B",4), "=B3"); u.Show = true;
In this code, I first create an ExcelHelper.Utilities object which provides the methods to create and manipulate the workbook. I then add a sheet, set specific cell contents, and show the end result. This is sufficient for assembling most simple workbooks.
The code for ExcelHelper is below. Note that we needed to import the Microsoft Excel Core Library (in the COM tab):
using System; using Microsoft.Office.Core; using System.Diagnostics;
publicbool Show { get { returnthis.excelApp.Visible; } set { this.excelApp.Visible = value; } }
} }
Notice that I also created a "Cell" object. This simply takes a value in the familiar Letter(Column)-Number(Row) format and translates it to a Number-Number used by Excel's automation object.
using System;namespace ExcelHelper{ publicstruct Cell { public Cell(int intRow, int intCol) { _intRow = intRow; _intCol = intCol; } public Cell(string strCol, int intRow) { _intRow = intRow; _intCol = ConvertStringToIntColumn(strCol); if (_intCol > 256) thrownew ArgumentException("Column cannot be greater than IV (256). Value was '" + strCol + "'."); } privatestaticint ConvertStringToIntColumn(string strCol) { strCol = strCol.ToUpper(); //A --> 1, BD --> (26+4), IV//Can only have max of two characters.char[] ach = strCol.ToCharArray(); if (ach.Length == 0) thrownew ArgumentException("Column cannot have 0 length. Must be A - IV. Was: '" + strCol + "'."); elseif (ach.Length == 1) return GetNumberValue(ach[0]); elsereturn 26*GetNumberValue(ach[0]) + GetNumberValue(ach[1]); } privatestaticint GetNumberValue(char c) { return (int)c - 64; } privateint _intRow, _intCol; publicint Row { get { returnthis._intRow; } } publicint Column { get { returnthis._intCol; } } }}
Change History: This post was updated on July 5, 2005. See Update at bottom.
You can't add two identical events to the same Html control. For example, if you have a button with an onclick event, and you add another onclick at the server, the second event will overwrite the first. This becomes a problem if you need to add an event to some variable control, but you don't know what events that control already has. However, .Net provides a way to handle this.
We can:
See if the variable html control already has an onclick event.
If it does, then dynamically register a new script that calls both the original onclick script as well as the new script.
Replace the variable control's onclick value with that new script.
The following code snippets this. This first block shows a simple JavaScript (with a parameter) being called by a button's onclick event.
function DoStuff1(var1) { alert('1: ' + var1); }
...
...
This snippet shows the server code to check if an onclick event already exists, and add a new wrapper if needed. Note that it persists the onclick values in viewstate to ensure that the wrapper function doesn't wrap itself upon postback. For example, if the first onclick event called DoStuff1(), and we wanted to dynamically add a new onclick function DoStuff2(), we could create a wrapper function Wrapper1() that called both functions, and was called from the onclick. Wrapper1() becomes the new value of the button's onclick.
private void Page_Load(object sender, System.EventArgs e) { string s = this.Button1.Attributes["onclick"]; if (!Page.IsPostBack) OriginalJSFunction = s; if (s == null) this.Button1.Attributes.Add("onclick","DoStuff2()"); else { if (!Page.IsClientScriptBlockRegistered("JS1")) Page.RegisterClientScriptBlock("JS1",GetJS(OriginalJSFunction)); this.Button1.Attributes.Add("onclick","Wrapper1()"); } }
private string OriginalJSFunction { get { object o = ViewState["OriginalJSFunction"]; if (o == null) return ""; else return o.ToString(); } set { ViewState["OriginalJSFunction"] = value; } }
private string GetJS(string strOriginalFn) //will handle initial JS with params { System.Text.StringBuilder sb = new System.Text.StringBuilder(); sb.Append(@"
");
return sb.ToString(); }
While this approach is tedious, it lets you dynamically add an event to a control without overwriting that control's existing event. This is useful when making your own custom controls that integrate with existing Html controls already on the page.
UPDATE July 5, 2005
Since writing this post, I've come across a cleaner technique. You can run multiple JavaScripts simply by listing them in the onclick event:
Therefore we can simplify our codebehind. This has two main changes: (1) It modifies OriginalJSFunction to append the ";" to the end of a non-function, (2) It then always adds an onclick method, as opposed to checking if there already exists an onclick method.
privatevoid Page_Load(object sender, System.EventArgs e) { string s = this.Button1.Attributes["onclick"]; if (!Page.IsPostBack) OriginalJSFunction = s;
