Scheduled Integration is hard. I remember being involved in projects where developers would get a copy of the latest source code and a task, and race off like horses at the track while they spent a few weeks implementing their assigned feature. At the end of these many weeks was a scheduled integration, where all developers in the team would reconvene with their code modifications, and try to get their respective code to play well in a single sandbox. Project managers always seemed to expect that this would be a quick and painless process, but, of course, it never was. Reintegration of the code sometimes took just as long as the implementation, itself. Even if developers are all working in one big room throughout the project, the code remains isolated, never shared or reintegrated. Fortunately, Continuous Integration specifically focuses on this problem by reducing (and often eliminating) the end-of-project reintegration cycle through a continuous constant integration process.

Continuously Integrating your Mind

For a moment, compare  the process of software development with the process of learning new technologies in your field. In the past few years of Microsoft's .NET platform, we've seen .NET 2.0, Windows Communication Foundation, Windows Presentation Foundation, Workflow Foundation, Extension Methods, Generics, Linq, Dynamic Data, ASP.NET MVC, and more. For a developer that has not kept up with the latest trends, to suddenly and quickly catch up with the bleeding edge would be a major undertaking, and psychologically intimidating. This Scheduled Integration approach to learning can be overwhelming; the learning process is isolated, and only occasionally reconnects to the latest technologies. However, a developer that has kept up with the trends, learning through a continuous integration process, has been constantly updating, constantly learning the Next Big Thing over these past few years; this developer is already long familiar with .NET 2.0, WF, WCF, WPF, Generics, and Linq, and is now working on only Dynamic Data and MVC. The Continuous Integration process ensures that those involved are current with the latest developments in their project, and that the overwhelming burden of integration at the end of a project is instead simplified by distributing it throughout the course of the timeline.

Core Continuous Integration

At its core, Continuous Integration, or CI, is a development model where developers regularly commit and update against their source control repository. By committing (giving everyone else access to your code changes) and updating (getting code changes from everyone else), the scheduled, tedious integration process at the end of the project is eliminated. Added on top of this single fundamental, is ensuring the code works: Automating the build through scripting frameworks like Make, MSBuild, or NAnt helps developers validate their status quickly, by invoking a validation step that not only compiles, but executes a suite of unit tests against the code base. Validation results, along with the latest valid code, are then made easily accessible to anyone on the team.

The ten tenets of Continuous Integration are:

1. Maintain a Source Code Repository
   Use a code repository, such as Subversion, Bazaar, or even Visual Source Safe. This allows a central point-of-truth for the development team, against which code can be committed or updated. This even applies to a development team consisting of only one person, as a local hard drive alone cannot provide change logs, revision differences, rollback capability, and other benefits of source management.
2. Commit Frequently
   As other developers commit code changes, your local version of the code will get further and further from the revision head, thus increasing the likelihood of a conflicting change that will need manual resolution. By committing often (and by association, updating even more often), everyone can stay very close to the revision head, reducing the likelihood of conflicts, and reducing time to integrate code and functionality.
3. Self-Testing Code
   "It compiles" is not sufficient criteria for determining if a project is ready for delivery or deployment to the client. Some level of testing must be conducted against each compile to measure application quality. Using unit tests, functional tests, or some other form of automated acceptance tests, the code can evaluate itself against expected outcome, providing a much more accurate and granular metric for readiness.
4. Automate the Build
   Using Make, NAnt, MSBuild, or similar frameworks, consolidate execution of your full build into a single command, or a single icon on your desktop. These scripts should execute a full compile, and run your full suite of testing and evaluation tools against the compile.
5. Build Fast, Fail Fast
   Even if your build is automated, no one wants to wait 30 minutes for the build to complete. Building your code should not just be a lunch-break activity. Keep the build fast to enable developers to do so as often as possible. And if there is a problem, fail immediately.
6. Build Every Mainline Commit on an Integration Machine
   We all have applications on our local desktops that will not be present in Production. Instant Messenger, iTunes, Visual Studio, and Office are all common for us, but rare in Production. However, these applications can conflict or distort build results, such as a reference to an Office assembly that is not included in your deployment package. By executing the automated build on an integration machine (iTunes free, and using a CI suite like Hudson or CruiseControl), you can increase confidence in your application and eliminate "it works on my box!"
7. Automate the Deployment
   Manual code deployment is a mundane, highly repetitive, error-prone, and time-consuming process that is ripe for automation. The time commitment adds to the stress when QA requests yet another deployment to the testing environment, particularly when all of the developers are in 80-hour-week crunch mode. In turn, this stress reduces deployment quality; environments often have configuration differences, such as different database connection strings, credentials, or web service URLs, and often only one configuration change needs to be overlooked to cause the entire system to malfunction. Automate this task, so that it can be executed easily, dependably, and often.
8. Test in a Clone of the Production Environment
   In addition to Office and iTunes not being a part of the production server build, there are aspects of the production environment that are not a part of the desktop environment. Server farms, federated databases, and load balancing are examples of things that do not exist on the developer desktop, but do exist in production, and can cause surprises if they are not considered during development and testing. Consider the haves and the have-nots in your test environment, and eliminate these surprises. And if the cost of another production environment is out of reach, consider Virtual Machines. VMs have significantly reduced the cost of creating a watered down test environment that still has things like server farms or database clusters; even if you cannot exactly replicate your production configuration, mitigate your risk by reducing the differences between your test and production environments.
9. Everyone Can View the Latest Build Results
   The underlying driver behind Continuous Integration is transparency and visibility. Communication enables both transparency and visibility by allowing everyone on the team to know the full status of the build. Did it compile? Did the unit tests pass? Which unit tests failed? Did the deployment work? How many seconds did it take to compile the application? Who broke the build? Continuous Integration suites, such as Hudson or CruiseControl, provide reporting mechanisms on build status. Make this status available to anyone, including project managers, and even the sales guy.
10. Everyone Can Get the Latest Executable
    On a software development project, communication is more than just green icons (successful builds) and red icons (failed builds). Communicate the ones and zeros, in the form of your compiled application, to your team. By allowing other developers, testers, and even the sales guy (perhaps for a demo) to get the latest bits for themselves, developers can focus on writing code.

Benefits of Continuous Integration

By continuously integrating all new code and feature sets, development teams can eliminate that long and tedious scheduled integration effort, which reduces overall effort, time line, and budget. Through self-testing code and building every mainline commit, code is continuously tested against a full suite of tests, allowing quick analysis and identification of breaking changes. Through the easy accessibility of the latest bits, the team can test early and often, allowing quick identification of broken functionality, and for early identification of features that don't quite align with what the client had in mind. And finally, the immediate and public feedback on the success or failure of the build provides incentives for developers to write code in smaller increments and perform more pre-commit testing, resulting in higher quality code.

I consider Continuous Integration to be an essential, required part of any development effort, every time. I started using CI in 2004, and I have since become dependent on it. I even have a Continuous Integration box at home, validating home projects for my development-team-of-one, and I am comforted by having a Continuous Integration server analyzing and validating every code change that I make. I break unit tests as often as anyone else does, and there still continues to be plenty of times that I even break the compile. At least once, every developer among us has checked in the project file while forgetting to add myNewClass.cs to Source Control. It will break the compile every time. Fortunately, Continuous Integration is always watching my code commits; it will let me know that it could not find myNewClass.cs, every time. And my application's quality is remarkably better for it. Every time.

Continuous Integration: It's more than just a toolset
Wednesday, 08 October, 2008 @ 6:00pm
SRT Solutions
206 South Fifth Ave, Suite 200
Ann Arbor, MI 48104

Session Abstract:

Continuous Integration: It's more than just a toolset
Thursday, 11 September, 2008 @ 6:00pm
New Horizons
4488 West Bristol Road
Flint, MI 48507

Session Abstract:

CruiseControl.Net 1.3 was released this morning. To me, most important was the new Build Queue functionality, stopping multiple projects from building at the same time. If ProjectB depends on ProjectA, and they both get code changes committed at the same time, they will fail from contention. Either ProjectA will fail because it can’t delete its old assemblies (because ProjectB has a lock on them) or ProjectB will fail because it can’t find the ProjectA assemblies (because ProjectA deleted them in its rebuild).

No more.

I’m so excited that I am already upgrading our servers!

For those that missed the announcement last week (like I did), the latest version of CruiseControl.Net has been released.

I plan on checking it out this week, then possibly upgrading our Build environment on Saturday. There are some modifications that I am really excited about:

• Log4Net is used. (Default: Rolling file appender for logging server output.) If the traditional Log4Net configuration block is included in the application configuration file, I will probably change that to the ADONet appender, instead.
• Users can volunteer to fix a broken build. How sweet is that!?!
• <prebuild /> section allows custom tasks to run prior to the build. This one is a big bonus; previously, if something went wrong with the build, often the external log files (NUnit, FXCop) from the previous build would get included in the current build’s report. Now the prebuild can give them the boot.
• Caching is used on WebDashboard. We have some huge log files and some not-so-powerful build servers. Sometimes it takes the machine a while to process the XSL. I am hoping that caching will help with that.
• WebDashboard can stop and start projects. I am very excited about the ability to pause individual projects without having to modify the setup or stop the entire service.

This seems like a nice package (Release Notes). I am eager to pull it down and give it a go.

One gotcha that everyone should be aware of: Old versions of the dashboard and CCTray are incompatible with the new version of the service, so both will need to be replaced. Give your development team a heads-up, so they know to replace their tray installation as soon as the new server version is installed and online.

NAnt hates .Net’s resource files, or .resx. Don’t get me wrong–it handles them just fine–but large quantities of resx will really bog it down.

Visual Studio loves resx. The IDE will automatically create a resource file for you when you open pages and controls in the ‘designer’ view. Back when we still used Visual SourceSafe as our SCM, Visual Studio happily checked the file in and forgot about it. Now, our 500+ page application has 500+ resource files. Most of these 500+ resource files contain zero resources, making them useless, pointless, and a detriment to the build.

This morning I went through the build log, noting every resx that contained zero resources, and deleted all of these useless files.

The compile time dropped by 5 minutes.

Moral of the story: Be weary of Visual Studio. With regards to resx, VS is a malware program that’s just filling your hard drive with junk. If you use resx, great, but if you don’t, delete them all. NAnt will love you for it.

CruiseControl .Net 1.0 has been released. download | release notes

This is a must upgrade for anyone running v0.9 or earlier. There are many updates that I am excited about, most notably the overhaul to CCTray (the client-side build monitoring tool that sits in your system tray). Our developers have had to use Firefox’s CC.Net monitor extension to monitor multiple builds, simultaneously. No more.

We will be upgrading within the next week.

MSIExec error code 1605 has been a thorn in my side for quite a while. When an MSI was command-line deployed by one user (manually deployed by me in the middle of the day), it couldn’t be uninstalled by another (automation during the nightly) due to the “Just Me” default. If I installed it through using the UI, and installed it for use by “Everyone”, then the nightly would build just fine. I needed a way to run an “Everyone” install from the command line, but Google wasn’t helping me out. Unfortunately, Microsoft does not seem to have a lot of documentation on this functionality, either.

It further frustrated me this morning when my nightlies were failing again, but only on one server. Of course, I manually deployed the package to this same server to a few days ago. I tried Google again, and this time hit pay dirt. Executing it with ALLUSERS=2 in the command line makes it available for everyone. Apparently, it forces an “Everyone” install for the UI, too.

Finally I can pull the thorn out.

MSIExec /i mypackage .msi … ALLUSERS=2

I know. I haven’t posted in a while. But I’ve been crazy busy. Twelve hour days are my norm, right now. But enough complaining; let’s get to the good stuff.

By now you know my love for PsExec. I discovered it when trying to find a way to add assemblies to a remote GAC [post]. I’ve found more love for it. Now, I can remotely execute my performance tests!

Execute LoadRunner test using NAnt via LoadRunner:

<exec basedir="${P1}"
  program="psexec"
  failonerror="false"
  commandline='\${P2} /u ${P3} /p ${P4} /i /w "${P5}" cmd /c wlrun -Run
    -InvokeAnalysis -TestPath "${P6}" -ResultLocation "${P7}"
    -ResultCleanName "${P8}"' />

(I’ve created generic parameter names so that you can read it a little better.)
P1: Local directory for PsExec
P2: LoadRunner Controller Server name
P3: LoadRunner Controller Server user username. I use an Admin-level ID here, since this ID also needs rights to capture Windows PerfMon metrics on my app servers.
P4: LoadRunner Controller Server user password
P5: Working directory on P2 for 'wlrun.exe', such as C:\Program Files\Mercury\Mercury LoadRunner\bin
P6: Path on P2 to the LoadRunner scenario file
P7: Directory on P2 that contains all results from every test
P8: Result Set name for this test run

'-InvokeAnalysis' will automatically execute LoadRunner analysis at test completion. If you properly configure your Analysis default template, Analysis will automatically generate the result set you want, save the Analysis session information, and create a HTML report of the results. Now, put IIS on your Controller machine, and VDir to the main results directory in P7, and you will have access to the HTML report within minutes after your test completes.

Other ideas:

• You can also hook it up to CruiseControl and have your CC.Net report include a link to the LR report.
• Create a nightly build in CC.Net that will compile your code, deploy it to your performance testing environment, and execute the performance test. When you get to work in the morning, you have a link to your full performance test report waiting in your inbox.

The catch for all of this: you need a session logged in to the LoadRunner controller box at all times. The '/i' in the PsExec command means that it interacts with the desktop.

Sidenote

PsExec is my favorite tool right now. I can do so many cool things. I admit, as a domain administrator, I also get a little malicious, sometimes. The other day I used PsExec to start up solitaire on a co-workers box, then razzed him for playing games on the clock.

With our new nightly database restore we now have the desire to automatically run all of the change scripts associated with a project. We’ve found a way; I created a NAnt script that will parse the Visual Studio Database Project (or "DBP") and execute all of the change scripts in it. Here’s how we got there.

Problem 1: Visual Studio Command Files are worthless

Our first idea was to have everyone update a command file in the DBP, and have NAnt run it every night. Visual Studio command files are great and all, but we have discovered a problem with them: they do not keep the files in order. We have named all of our folders (01 DDL, 02 DML, etc) and our change scripts (0001 Create MyTable.sql, 0002 AddInfoColumn to MyTable.sql) accordingly so that they should run in order. We have found that the command file feature of VS.Net 2003 does not keep them in order but rather seems to sort them first by extension, then by order, or some similar oddness. Obviously, if I try to at InfoColumn to MyTable before MyTable exists, I’m going to have a problem. So, the command file idea was axed.

Problem 2: Visual SourceSafe contents can’t be trusted

Our second idea was to VSSGET the DBP directory in VSS and execute every script in it. However, the VSS store cannot be trusted. If a developer creates a script in VS.Net called ‘0001 Crate MyTable.sql’ and checks it in to the project, then proceeds to correct the spelling error in VS.Net to ‘0001 Create MyTable.sql’, VS does not rename the old file in VSS. Instead, it removes the old file from the project, renames it locally, then adds the new name to the project and to VSS. It also never deletes the old file name from the VSS store. Now, both files (’0001 Crate MyTable.sql’ and ‘0001 Create MyTable.sql’) exist in VSS. Performing a VSSGET and executing all scripts will run both scripts, which could lead to more troubles.

So, we can’t use a command file, because it won’t maintain the order. We can’t trust VSS, since it can have obsolete files. We can only trust the project, but how do we get a list of files, ourselves?

Fortunately, DBP files are just text in a weird XML-wannabe format. The NAnt script will open the file and run through it looking for every ‘SCRIPT’ entry in the file. If it finds a ‘BEGIN something’ entry, it assumes that ’something’ is a folder name, and appends it to the working path until it finds ‘END’, at which time it returns to the parent directory.

It’s not perfect. It still runs in to some problems, but here it is in v0.1 form.

<project name="RunDBPScripts" default="RunScripts">
<!–-
Execute all scripts in a VS.Net DBP
Author: Jay Harris, http://www.cptloadtest.com, (c) 2005 Jason Harris
License: This work is licensed under a  
   Creative Commons Attribution 3.0 United States License.  
   http://creativecommons.org/licenses/by/3.0/us/ 

This script is offered as-is.
I am not responsible for any misfortunes that may arise from its use.
Use at your own risk.
-–>
<!-– Project: The path of the DBP file –->
<property name="project" value="Scripts.dbp" overwrite="false" />
<!-– Server: The machine name of the Database Server –->
<property name="server" value="localhost" overwrite="false" />
<!-– Database: The database that the scripts will be run against –->
<property name="database" value="Northwind" overwrite="false" />
<target name="RunScripts">
        <property name="currentpath"
            value="${directory::get-parent-directory(project)}" />
        <foreach item="Line" property="ProjectLineItem" in="${project}">
            <if test="${string::contains(ProjectLineItem, 'Begin Folder = ')}">
                <regex pattern="Folder = &quot;(?’ProjectFolder’.*)&quot;$"
                    input="${string::trim(ProjectLineItem)}" />
                <property name="currentpath"
                    value="${path::combine(currentpath, ProjectFolder)}" />
            </if>
            <if test="${string::contains(ProjectLineItem, 'Script = ')}">
                <regex pattern="Script = &quot;(?’ScriptName’.*)&quot;$"
                    input="${string::trim(ProjectLineItem)}" />
                <echo message="Executing Change Script (${server+"\"+database}): ${path::combine(currentpath, ScriptName)}" />
                <exec workingdir="${currentpath}" program="osql"
                    basedir="C:\Program Files\Microsoft SQL Server\80\Tools\Binn"
                    commandline=’-S ${server} -d ${database} -i “${ScriptName}" -n -E -b’ />
            </if>
            <if test="${string::trim(ProjectLineItem) == 'End’}">
                <property name="currentpath"
                    value="${directory::get-parent-directory(currentpath)}" />
            </if>
        </foreach>
    </target>
</project>

I used an <EXEC> NAnt task rather than <SQL>. I found that a lot of the scripts would not execute in the SQL task because of their design. VS Command Files use OSQL, so that’s what I used. I guess those command files were worth something after all.

If you know of a better way, or have any suggestions or comments, please let me know.