I must admit that I don’t much care for PowerShell's default behaviour with respect to errors, which is to continue on error. It feels very VB6 “On Error Resume Next”-ish. Given that it is a shell scripting language, I can understand why the PowerShell team chose this as a default. Fortunately you can change the default by setting $ErrorActionPreference = ‘Stop’, which terminates execution by throwing an exception. (The default value is Continue, which means the script prints the error and continues executing.) Unfortunately this only works for PowerShell commands and not external executables that return non-zero error codes. (In the shell world, a return code of zero (0) indicates success and anything else indicates failure.)

Take the following simple script:

'Starting script...'
$ErrorActionPreference = 'Stop'
ping -badoption
"Last Exit Code was: $LastExitCode"
rm nonexistent.txt
'Finished script'

Notice how execution continued after the ping command failed with an exit code of one (1) even though we have $ErrorActionPreference set to ‘Stop’. Also notice that the rm command, which is an alias for the PowerShell command, Remove-Item, did cause execution to abort as expected and ‘Finished script’ was never printed to the console. The discrepancy in error handling between PowerShell commands and executables is annoying and forces us to constantly think about what we’re calling – a PowerShell command or an executable. The obvious solution is:

'Starting script...'
$ErrorActionPreference = 'Stop'
ping -badoption
if
($LastExitCode -ne 0) { throw 'An error has occured...' } rm nonexistent.txt
'Finished script'

The error handling code adds a lot of noise, IMHO, and feels like a throwback to COM and HRESULTs. Can we do better? Jorge Matos, one of the psake contributors came up with this elegant helper function:

function Exec([scriptblock]$cmd, [string]$errorMessage = "Error executing command: " + $cmd) { 
  & $cmd 
  if ($LastExitCode -ne 0) {
    throw $errorMessage 
  } 
}

Note the “& $cmd” syntax. $cmd is a scriptblock and & is used to execute the scriptblock. We can now re-write our original script as follows. (N.B. Exec function is elided for brevity.)

'Starting script...'
$ErrorActionPreference = 'Stop'
exec { ping -badoption }
rm nonexistent.txt
'Finished script'

The script now terminates when the bad ping command is executed. We do have to remember to surround executables with exec {}, but this is less noise IMHO than having to check $LastExitCode and throwing an exception.

For those of you using psake for your builds, the Exec helper function is included in the latest versions of the psake module. So you can use it in your build tasks to ensure that you don’t try to run unit tests if msbuild fails horribly.

Happy Scripting!

As many of you know, I am an independent consultant and use my own laptop when possible. I’ve got all my tools set up the way I like them and everything else that I need to be productive. Given that I work for multiple clients, I can’t join my laptop to any particular client’s domain. First is the hassle factor, especially when switching between different clients within a week. Each domain join requires a domain admin to authorize the join by typing in his/her credentials when prompted on my laptop. Second I don’t want a client’s Group Policy being applied to my laptop. Third – and more importantly – is the non-disclosure agreements that I sign with clients. If I join my laptop to a domain, the domain admins have full rights to my machine and hence data from other clients. So domain joining just isn’t an option.

In most cases, not being joined to a client’s domain doesn’t make one iota of difference. You need to access a network share or printer, browser to it and you will be prompted for domain credentials. The fact that you’re using different domain credentials to access the resource from those that you logged in with doesn’t matter one bit. If you want to expedite the process and not wait for an authentication time-out, you can utilize NET USE from the command line to tell Windows which credentials you want to use when accessing certain computers. You can even make them persistent or roll the whole thing into a batch script that you can execute whenever at a particular client.

net use \\server /user:domain\username /persistent:yes

Unfortunately this doesn’t work in all cases. One of my longstanding development pet peeves has been certain tools – I’m looking at you SQL Server Management Studio and SQL Query Analyzer – that don’t allow you to specify alternate domain credentials for authentication. For example, SQL Server Management Studio allows you to log into a SQL Server instance using Windows Authentication or SQL Server Authentication. If the SQL instance requires Windows Authentication – the recommended configuration – SQL Server Management Studio uses your logged in credentials. This works well if your computer is part of the domain, but fails horribly if not. It doesn’t let you specify alternate credentials or even prompt you for alternate credentials if the log-in fails.

I’ve tried seemingly everything. NET USE doesn’t help here because NET USE is specifically for network shares.

net use \\sql-server-name /user:domain\username # DOES NOT WORK - Only provides the domain credentials when accessing shares

RUNAS also fails – either the SHIFT right-click variety or command line – as it tries to run the command locally as the domain user, who is unknown by your computer because you’re not part of the domain.

runas /user:domain\username “C:\Program Files (x86)\Microsoft SQL Server\100\Tools\Binn\VSShell\Common7\IDE\Ssms.exe”

For years (yes, years) I have resorted to using Remote Desktop to log into a domain computer so that I could run SQL Server Management Studio, used a domain-joined virtual machine, or begged co-workers to run commands for me. Now I feel foolish because I stumbled upon a solution that has been built into Windows for years. It is a simple command line switch for the RUNAS command that I never noticed: /netonly. (Note that the /netonly flag is not accessible via the SHIFT right-click menu, only via the command line.)

runas /netonly /user:domain\username “C:\Program Files (x86)\Microsoft SQL Server\100\Tools\Binn\VSShell\Common7\IDE\Ssms.exe”

Success! And SQL Server Management Studio running using /netonly domain credentials. The command is run as my local user, but uses the supplied domain credentials only when accessing the network.

I can access remote SQL Servers using Windows Authentication without problem now! (You’ll have to take my word for it or try it yourself as it would be impolite for me to show screenshots of me accessing a client’s SQL Server.) Hopefully this makes some other consultant’s life a little bit easier.

The Edmonton Code Camp was a lot of fun. It was wonderful to catch up with friends – old and new – over lunch and dinner. Thanks to Dave Woods for inviting me.

I have been enjoying sharing my ideas about convention-over-configuration and how it can simplify software development. You expend some serious brain power over figuring out how to enable your application-specific conventions, but everything after that flows easily and without repetition. You end up doing more with less code. During the talk, I demonstrated how frameworks like Fluent NHibernate, AutoMapper, Castle Windsor, ASP.NET MVC, and jQuery support this style of development. (Links below.) I only scratched the surface though. Other frameworks like StructureMap and FubuMVC also are heavily convention-based. With a bit of creative thinking, you can use these techniques in your own code to reduce duplication and increase flexibility.

• Fluent NHibernate
• NHForge
• AutoMapper
• Castle Windsor
• ASP.NET MVC
• jQuery & jQuery UI

For those of you who attended, you’ll realize why Darth Vader accompanies this post. For everyone else, you’ll have to check out the slidedeck and code:

PPTX | Code

At first glance, PowerShell appears to be yet another command shell with the interesting twist that you pipe objects between commands rather than strings. But there is more to PowerShell than that. One fascinating area is PowerShell Providers. (PowerShell Providers aren’t anything new as they’ve been there since v1. So I’m not the first – nor will I be the last – to blog about them, but hopefully some folks starting out with PowerShell find this useful…)

We’ll start with a simple example using “ls” to list the contents of a directory:

ls c:\

Now “ls” is just a two-letter version of “dir” and both are aliases for “Get-ChildItem”. How do I know that?

ls alias:

This prints out all current aliases. That funky “alias:” is a PowerShell provider. If you want a specific alias, you can “ls alias:ls” or “ls alias:dir”.

To get a list of currently installed PowerShell providers, you can use Get-PSProvider:

You should notice a few interesting entries there. You want a list of environment variables? The Environment provider does the trick:

Note that providers aren’t read-only. Let’s say you want to temporarily add a directory to your path. In cmd.exe, you would do the following:

set PATH = %PATH%;<EXTRA_DIR>;

In PowerShell, you use the Environment provider:

$env:PATH += ";<EXTRA_DIR>";"

Notice the env: prefix that tells PowerShell that the variable is handled by the Environment provider. (Notice above that “Env” is listed as the “drive” for the Environment provider.)

Let’s explore some more…

Notice that changing drive letters is actually handled by the Function provider and are just commands.

It gets more interesting with the Registry provider through which you can access HKEY_LOCAL_MACHINE via hklm: and HKEY_CURRENT_USER via hkcu:.

You even get tab completion while typing. (Try ls hkcu:<TAB><TAB><TAB> to see various subkeys for HKEY_CURRENT_USER.) And assuming that you have write permission to the registry keys, you can set them too!

PowerShell providers aren’t limited to those shipped by Microsoft. You can in fact write your own, though I’ve never tried it. People have written their own providers for everything from SharePoint to Subversion.

So go check out PowerShell providers. Happy Scripting!

A few weeks ago Richard and Carl invited me to appear on .NET Rocks again and I jumped at the chance. I had a great time talking to them about doing more with less (writing less, but smarter code) and how convention-over-configuration changes the way that we develop software for the better. Check it out .NET Rocks #475 featuring yours truly!

Thanks to everyone who came out to my presentation last night at the Calgary .NET User Group. I enjoyed talking using convention-over-configuration techniques for doing more with less code. I demonstrated how frameworks like Fluent NHibernate, AutoMapper, Castle Windsor, ASP.NET MVC, and jQuery support this style of development. (Links below.) I only scratched the surface though. Other frameworks like StructureMap and FubuMVC also are heavily convention-based. With a bit of creative thinking, you can use these techniques in your own code to reduce duplication and increase flexibility.

• Fluent NHibernate
• NHForge
• AutoMapper
• Castle Windsor
• ASP.NET MVC
• jQuery & jQuery UI

For those of you who attended, you’ll realize why Darth Vader accompanies this post. For everyone else, you’ll have to check out the slidedeck and code:

PPTX | Code

I’ll be presenting at the Calgary .NET User Group next week. Come out for a fun discussion and lively discussion on improving your application development using convention-over-configuration techniques.

Abstract

As developers, we spend an inordinate amount of time writing "glue code". We write code to transform database rows to domain objects... domain objects to view-models or DTOs... We write code to configure inversion of control containers and wire dependencies together. We write code to style our UIs and respond to UI events. Wouldn't it be nice if this could happen automagically for us? This session will look at using convention-based approaches using Fluent NHibernate, AutoMapper, Castle Windsor, and jQuery to reduce the amount of repetitive code and accelerate application development.