memory.Refresh()

Using the latest version (1.35) of FxCop, it is possible to exclude generated warning messages in the source code instead of having to exclude them in the FxCop project file. This is accomplished using the System.Diagnostics.CodeAnalysis.SuppressMessageAttribute class.

To exclude a message, simply mark up the method with the SuppressMessage attribute, declaring both the rule category and the specific rule to exclude:

[SuppressMessage("Microsoft.Design", "CA1062:ValidateArgumentsOfPublicMethods")]
public string ConvertToString(object obj)
{
    return obj.ToString();
}

The above example will exclude the “Validate Arguments Of Public Methods” rule from the Design category for the ConvertToString method (although in this contrived example it is probably a bad idea to do so as passing in a null will clearly cause problems).

One extra “tweak” that can be utilised in this scenario is the Justification property. Altering the above code to:

[SuppressMessage("Microsoft.Design", "CA1062:ValidateArgumentsOfPublicMethods", Justification="I do have a valid reason")]
public string ConvertToString(object obj)
{
    return obj.ToString();
}

will allow the person excluding a message to provide a reason for doing so in the code, alongside the exclusion. FxCop (v1.35) currently doesn’t display this in its output, but will do in the next release (source). It does however output the justification to a generated report if the options are set to output exclusions to the report.

Excluding FxCop messages in the source code has advantages over excluding them in the FxCop project as it demonstrates that the message has been specifically excluded for that particular case, but also will withstand class and namespace changes. It also makes switching from the standalone FxCop to Visual Studio’s code analysis an easier process.

In order to allow the SuppressMessage attribute to work, a CODE_ANALYSIS conditional compilation symbol must be defined for the project. Without this, FxCop will ignore the suppressed attribute and will still generate a warning.

More details can be found on the FxCop blog.

When remoting, it’s possible to get a strange exception if a SQLException occurs on the server.

The stack trace of the error is:

Exception: System.Runtime.Serialization.SerializationException
Message: Member name ‘System.Data.SqlClient.SqlError server’ not found.

Stack trace:
at System.Runtime.Serialization.Formatters.Binary.ReadObjectInfo.GetMemberTypes (String[] inMemberNames)
at System.Runtime.Serialization.Formatters.Binary.ObjectMap..ctor(String objectName, String[] memberNames, BinaryTypeEnum[] binaryTypeEnumA, Object[] typeInformationA, Int32[] memberAssemIds, ObjectReader objectReader, Int32 objectId, BinaryAssemblyInfo assemblyInfo, SizedArray assemIdToAssemblyTable)
at System.Runtime.Serialization.Formatters.Binary.ObjectMap.Create(String name, String[] memberNames, BinaryTypeEnum[] binaryTypeEnumA, Object[] typeInformationA, Int32[] memberAssemIds, ObjectReader objectReader, Int32 objectId, BinaryAssemblyInfo assemblyInfo, SizedArray assemIdToAssemblyTable)
at System.Runtime.Serialization.Formatters.Binary.__BinaryParser.ReadObjectWithMapTyped(BinaryObjectWithMapTyped record)
at System.Runtime.Serialization.Formatters.Binary.__BinaryParser.ReadObjectWithMapTyped(BinaryHeaderEnum binaryHeaderEnum)
at System.Runtime.Serialization.Formatters.Binary.__BinaryParser.Run()
at System.Runtime.Serialization.Formatters.Binary.ObjectReader.Deserialize(HeaderHandler handler, __BinaryParser serParser, Boolean fCheck, IMethodCallMessage methodCallMessage)
at System.Runtime.Serialization.Formatters.Binary.BinaryFormatter.Deserialize (Stream serializationStream, HeaderHandler handler, Boolean fCheck, IMethodCallMessage methodCallMessage)
at System.Runtime.Remoting.Channels.CoreChannel.DeserializeBinaryResponseMessage(Stream inputStream, IMethodCallMessage reqMsg, Boolean bStrictBinding)
at System.Runtime.Remoting.Channels.BinaryClientFormatterSink.SyncProcessMessage(IMessage msg)

What is actually happenning here is that the System.Data.dll on the client is a slightly different version to that on the server. When the SQLException is deserialised at the client end, there is a missing property on the client that cannot be deserialised.

This error usually occurs when remoting to a Windows 2003 Server box from a non-Windows 2003 client box. The client has a System.Data.dll of version 1.1.4322.2032. The server has a System.Data.dll of version 1.1.4322.2300. The difference is that the server property of the SqlError class is never set in v1.1.4322.2032, causing the serialisation error. More detailed info can be found on the DevNewsGroups site.

Microsoft has two knowledge base articles, KB884871 and KB887549, that pertain to this issue. The suggested solution is a .NET Framework 1.1 post-SP1 hotfix, but this is only available by contacting Microsoft directly.

The hotfix solves the problem, but then stops old SqlError types from being deserialised, resulting in the exception:

Exception: System.Runtime.Serialization.SerializationException
Message: Wrong number of Members. Object System.Data.SqlClient.SqlError has 8 members, number of members deserialized is 7.

Stack trace:
at System.Runtime.Serialization.Formatters.Soap.ReadObjectInfo.PopulateObjectMembers()
at System.Runtime.Serialization.Formatters.Soap.ObjectReader.ParseObjectEnd(ParseRecord pr)
at System.Runtime.Serialization.Formatters.Soap.ObjectReader.Parse(ParseRecord pr)
at System.Runtime.Serialization.Formatters.Soap.SoapHandler.EndElement(String prefix, String name, String urn)
at System.Runtime.Serialization.Formatters.Soap.SoapParser.ParseXml()
at System.Runtime.Serialization.Formatters.Soap.SoapParser.Run()
at System.Runtime.Serialization.Formatters.Soap.ObjectReader.Deserialize(HeaderHandler handler, ISerParser serParser)
at System.Runtime.Serialization.Formatters.Soap.SoapFormatter.Deserialize(Stream serializationStream, HeaderHandler handler)
at System.Runtime.Serialization.Formatters.Soap.SoapFormatter.Deserialize(Stream serializationStream)

This means that the hotfix is not backwards-compatible. The only way to get everything happy is to upgrade all Windows 2003 boxes to SP1 and all clients (Windows 2000 and Windows XP) to .Net 1.1 SP1 plus the hotfix.

Microsoft recently released Guidance Explorer, a tool that contains a browsable collection of best-practice patterns for developing .Net and ASP.Net applications.

Both the tool and its guidance library have regular updates, containing best-practices for performance and security related issues.

The tool also has the ability to add custom sets of guidance to allow corporate/team standards to be included.

Andras Belokosztolszki from Red Gate posted an interesting article about the sp_rename stored procedure in SQL Server and the pitfalls that can occur after its use.

When a stored procedure is created, an object is created in the sysobjects table (or the sys.objects view in SQL Server 2005), and the textual definition is stored in the syscomments table (or the sys.sql_modules view in SQL Server 2005).

When using sp_rename to rename a stored procudure, the definition is left intact and only the name is changed in the sysobjects table (or sys.objects view). This means that the definition of the stored procedure stored in database now has the wrong name.

The best way of renaming a stored procedure is to completely delete it and then recreate it with the new name.

The following SQL displays the definitions of all stored procedures in a database alongside their names:

SQL Server 2000

SQL Server 2000
SELECT V2.[name], V1.[text]
    FROM syscomments AS V1, sysobjects AS V2
    WHERE V1.[id] = V2.[id]
    AND (V2.[xtype] = 'P' OR V2.[xtype] = 'F')

SQL Server 2005

SQL Server 2005
SELECT V2.[name], V1.[definition]
    FROM sys.sql_modules AS V1, sys.objects AS V2
    WHERE V1.[object_id] = V2.[object_id]

It is worth noting that enterprise manager uses the sp_rename stored procedure when using the right-click rename option (although SQL Server 2005 management studio does try to correct this problem if you view the definition of a renamed stored procedure by replacing the original stored name with the new one).

Aaron Stebner posted a recent article about the available installation modes for the .Net Framework 2.0.

There are three supported modes:

Standard Mode

All UI screens are displayed, including warnings about missing prerequisites, existing beta versions, errors and reboots.

The setup includes a multi-lingual UI that chooses the correct language from the user’s operating system UI language settings, so a user should always see the install in their preferred language.

Unattended Mode

All UI screens are suppressed, except for a progress dialog during installation.

To install in unattended mode, use the following command line:

dotnetfx.exe /q:a /c:"install.exe /qb"

Changing /qb to /qb! will hide the cancel button on the progress dialog, stopping the user from aborting the install.

Silent Mode

All UI screens are suppressed. This is useful for a custom installer to install the .Net Framework.

To install in silent mode, use the following command line:

dotnetfx.exe /q:a /c:"install.exe /q"

If using silent mode, any errors, previous beta versions or reboots must be handled by the hosting installer since the .Net Framework installer cannot display any dialogs to prompt the user. Aaron also provides details of possible return codes from the installer, how to detect previous beta versions of the framework and how to manage reboots.