Introducing myself into Deep Learning

Overview

It has been some time since my last blog post. Actually, it has been more than two years. The main reason is that something changed in my life when I started to train a neural network. Her name is Alexa.

Just to avoid confusion let me clarify that I am not a member of Amazon Alexa team. Alexa is the female version of my own name and it is the name of my little girl ;)

She is one of the reasons for this deep learning journey.

How did the journey begin?

Every journey has a motivation and this one is not the exception. It started on September 19 of 2016 when I carried her for the very first time. After saw her for a while I asked myself: How is possible that she can learn something in the future?

Months passed, and I saw her learning so fast effortless without too much “computational power” (apparently). She learned to walk, to dance, to almost talk, to solve a puzzle, to play basketball, to solve pyramid-piling rings puzzle, and even to scramble a Rubik’s cube ;)

Solving puzzle

Defending vs. Elizabeth

Scrambling a Rubik's cube

Beyond the obvious answers that she will learn by design, there is a lot of trial and error in her learning process. Several attempts to look for the best fit before she can learn something. I love to see her “computing the distance” from the expected result and her attempts at solving the pyramid-piling rings puzzle by removing a wrong ring and replacing it with the right one.

Dad is trying to build something
with blocks but I'm interested
in the neighbor’s dog ;)

What really happened is that her learning process motivated me to explore something “new”. A discipline that is called to be (if it is not really is) the new toolset for every single software developer. It turns out that the new hobby came with a practitioner approach but it required to be found first.

What happened next?

At this point, I started to watch the machine learning course videos of Andrew Ng using alternatives sources. Cubans (that live in Cuba) are not able to access the certification program at Coursera. In some way, the USA embargo to Cuba – (specifically the USA exportation laws) – also affects the deep learning global democratization process.

However, it doesn't worry me too much, actually never does. I can't get the certification but I can get the knowledge. Andrew’s course is actually motivating and didactically insuperable. It was able to bring back to life some math and algebra that I thought was dead in my brain and made me felt very comfortable implementing a vectorized version of the Stochastic Gradient Descent algorithm with Octave.

"We apologize for the inconvenience" said a @coursera message because also #block us, https://t.co/vsmxsoaMop #Cuba @cubavsbloqueo pic.twitter.com/4JpDtas3PH

— Alexánder Fernández (@alexfdezsauco) August 9, 2017

After understand how this works (almost just like Andrew used to say), and what kind of problems were possible to solve, just I wanted to put this in practice at the production level. Some researchers (friends of mine) sold me Tensorflow as the Holy Grail but I had some doubts about Python performance (still have).

The new hobby comes with the
practitioner approach

After some research, I found exactly what I was looking for. A deep learning library for JVM.  Deeplearnig4j (DL4j) is an excellent library and comes with this excellent book Deep Learning: A Practitioner's Approach from Adam Gibson and Josh Patterson. I just needed to read the preface, to identify me as a deep learning practitioner. It wasn't to hard notice that was the right book for me. I'm pretty sure that is also the right book for you.

DL4j also comes with a lot of helpful features and tools to assist the training process including Training UI, datavec, early stopping, even GPU support, and more, but we can talk about this in forthcoming posts.

DL4J Performance - examples/sec on Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz (No GPU required in this case)

Nice normal distribution shape for weights in the Layer Parameters Histogram (So, no regularization issues)

Some network details

Recently, I also started to work in build a proof of concept of an anomaly detection system built on top of DL4j, specifically by using Autoencoder networks with promising results but I can't give you anything in advance yet (just wait for it).

Autoencoders are supported by DL4j

Conclusions

I have a lot to learn about deep learning, but the journey has already begun and I have the intention to share it with you.

Btw, if you are not motivated to go “deep” with machine learning yet, be a father (or mother) first, and then let me know. I'm sure you will find the same "biological" inspiration when you witness how to "train a neural network" looks like ;)

Alexa talking with a large predator cat of stone

Introducing SharePoint Package Manager

Introduction

Half a year ago, I announced – in this post – the forthcoming release of a new secret weapon.  I apologize for the delay, but I entertained myself doing other things and I couldn't find a way to spend a couple of days on formulating it.

I could even tell you: ‘I had no time to do it’, but I learned the lesson – maybe I'm still learning – from my father. He always insisted me: ‘Alex, the time is the same’. So, what actually happened was a matter of time management and prioritization.

The true is, I also expected that something like this already exists at this time, but it doesn’t. So, I recalled this, wrote the minimal working code, push the sources and here it is.

What is SharePoint Package Manager? 

SharePoint Package Manager

The SharePoint Package Manager is a NuGet-based distribution and deployment system for SharePoint's solutions.

Basically, it's an extension of the SharePoint Central Administration site that automates the process for install or update solution packages from package sources (regular NuGet repositories).

Why NuGet?

Why not? NuGet is the widely accepted and popular package manager among .net developers.

There are also a lot of custom solutions and initiatives that use NuGet as the backend, for instance, ReSharper extension manager, OctopusDeploy, Chocolatey, Squirrel, you know, even Catel extending the modularity options of Prism.


There are only a few rules in order to use this package manager, starting with this 'new' concept: solution package.

What is a solution package?

A solution package is a regular NuGet package with its name ending in '.wsp' and with following structure:

A sample of solution package for SignalR.

Notice how the folder 'content' contains a wsp file with the same name of the solution package and that’s it.

A solution package can also have declared dependencies, but only between solution packages.

Maybe the naming convention looks weak, but there is only one package in the gallery with name Digst.OioIdws.Wsp, so, for me, is enough. Eventually, the package manager could also track non-solution packages through an ignore list.

Solution packages from NuGet Gallery. The Digst.OioIdws.Wsp isn't a solution package.

Managing package sources

The SharePoint Package Manager includes a page to manage the package source. You are able to add, remove, edit, enable or disable a package source.

Creating a new package source

The default package source is the NuGet Gallery.

Installing or updating solution packages

The SharePoint Package Manager also includes a page to manage the farm packages. Looking for the installed solutions in the SharePoint solution storage and making a join with available solution packages from the package sources, the page shows the available solution packages to install or update.

Managing the farm packages

Sorry about the name of the solution package in Spanish on the image above but was client choice and the picture was taken from real life scenario.

The SharePoint Package Manager provides an option to install or update solution packages. After clicking the button – on the right of the solution package info – the system will schedule a job to install or update the last available version of the selected solution package. If the selected package has dependencies, the package manager will also install or update all dependencies in the right order.

Tracking the install or update process.

What's next?

This is almost a draft or a proof of concept, remember just the minimal working code. So, you can try. I also published the SignalR.SharePoint.WSP solution package at NuGet gallery, so, you should see it as an available solution package.

Enjoy it, and let me know what you think. Even better, we can do this together, just fork the source on GitHub. There are a lot of things to do, including better user interface, REST services, a better approach for settings storage, ignore list, performance issues, msbuild extension, Visual Studio extension, and so on. You tell me.

In the meantime, I’m already working to turn more Catel’s rumors in true.

Talking about rumors, if everything goes right, I might update my resume in about seven months. Yes, your assumption is correct, I'm a father candidate ;).

'The Force Awakens' or How awake the communication channels on a development team?

Introduction

I’m actually not a follower of the Star Wars saga even when I saw all the seven movies – the last one as a cinema pirate copy – plus some episodes of animated series and also including some TV movies with the Ewoks when I was younger.

An Ewok

Who didn’t heard the story about Anakin Skywalker? The boy that was trained as Jedi by Obi-Wan Kenobi who eventually was turned – by his trend to the dark side of the force  and also influenced by a Sith – into Darth Vader, who also was the father of Luke – the new hope to get the balance of the force – and the princess Leia and so on.

Probably I missed something, some of these movies are too long and actually, some characters make me sick. But the true is that these movies were a turning point in the world of the visual effects.

Star Wars - The Force Awakens poster

I won’t talk about the movies but the title of the seven episode “The Force Awakens” inspired me to write a new post. 

The concept of the force, such energy field with source even genetic - cell with something call “midi-chlorian” - that empower the jedies knights  to move objects with his mind, get great reflex and anticipation, and also great skills as pilots, is actually unreal but in some scenarios could reveal its presence. 

Let me show you the way you can found an expression of the force as a result of an approach that allow you awake the team communication channel from a hypothetical scenario. 

Common communication channels

On software development teams the communication is a key factor. Not a few agile practice talk about it including informative workspace, release the plan, sit together, pair programming, ubiquitous language or even poker planning.  I have no doubt that every single agile practice is about to improve the communication in the team.

From a real life scenario communication channels could look like this way:

Common communication channels

The client tells to the product owner what he wants. The product owner can create a vision of the final product – optionally with a consultant - and share the expected result to the project lead in terms of requirements and also prioritize such requirements. Then project lead and the team should be able to create a roadmap in term of features and small increments to cover all client’s expectation.

The project lead must share systematically all his knowledge or the vision with the team looking for an effective and integrated solution. Notice that the project lead should be the only communication channel between each developer and the people that really knows what the client wants. 

But the world isn’t perfect and the project lead also should be ready to manage the changes on requirements including say NO or defers some requirements. There is a single rule, every single decision must be taken in collective with the team.

Actually, as a developer, I always expect an effective communication system. It could be done through public the iteration plan, plus daily meetings, retrospective, and demos, everything that allow the whole team get an integrated view of the thing that they are building. An informative workspace where they can share the issues or impediments and, of course, enjoy the progress.

Such approach should work and actually does.

But sometimes something happens that drive the team to a complete failure of the communication system.

Breaking the communication channels

Allow me to twist a bit the previous scenario.

Assume that suddenly the product owner and the external consultant start to manage the team in a very odd way. They do direct tasks and components assignments in private instead thought the project lead. From another side, but not less important the project lead always saying YES to each the product owner request without team consulting. At some point, the product lead also stops to communicate with the team at all, probably just a few emails or comments about some new client expectation, but not in real team dynamics.

The issue here is about product owner and consultant vision. They don't want to know about small increments. They want the final solution and production ready now. They don't know about the team dynamics or how they manage the source or about components integration. They also expect for some magical entity do the coordination work. But such thing doesn’t exist in software development.

The result: a completely broken communication system.

Broken communication channels

Awaking the communication channels

Could be spent some time before you really notice the disaster. The bug corrections lead to other incongruousness or to regressions. Even worse, people sitting together can't found a real way to coordinate their work. But something remains immutable, no one talk about the system beyond a few individuals' virtual chatting sessions. 

At this point, some practices could save them all. So, also assume that the team have a full dependency in a build automated system (just a build server), and have a well configured a deployment pipeline that run unit and integration tests and also run the deployment routines into quality assurance servers (including product owner's server).  

You as team member could ask yourself: Why don't use this to trigger the communication back?

So, you as a team member with integration needs can start to write a lot of integration test. For instance, for each disagreement, for each incongruousness, every assumption, every of non-final decisions, everything that anyone told aloud, you can translate it into an executable code that immediately will run on the build server, and eventually this approach will lead to stopping the deployment pipeline due by the tons of errors.

This is a point when a sort of energy field (a.k.a 'the force') should start to play its role. The product owner needs upgrades. But the upgrades are locked because the deployment pipeline is locked. So, eventually, someone would need to start talking again.

Conclusion

There aren’t nothing more effective to improve – not only to awake – the communication in your team than the integration tests. That is because coding is actually the communication channel that never dies. Even better if you really use practice like continuous integration you are already empowered – without “midi-chlorian” – to awake you communication channels at any time.

By the way, while I´m writing this lines, I noticed that Multivision channel - on Cuba - started to broadcast the saga again. So, if you never saw it, it's a good timing ;)

Star Wars Episode I: The Phantom Menace at Multivision

Remember: Your focus determines your reality. 
May the force be with you.

Simplest way to implement a state machine approach for SharePoint list items

Introduction

As you can read in Wikipedia “A finite-state machine (FSM) or finite-state automaton (plural: automata), or simply a state machine, is a mathematical model of computation used to design both computer programs and sequential logic circuits. It is conceived as an abstract machine that can be in one of a finite number of states. The machine is in only one state at a time; the state it is in at any given time is called the current state. It can change from one state to another when initiated by a triggering event or condition; this is called a transition. A particular FSM is defined by a list of its states, and the triggering condition for each transition.”

SharePoint developers frequently face to state machine problems and the typical solution includes a workflow implementation. Actually, SharePoint includes default workflows for common scenarios, for instance Approval (route a document or item for approval or rejection), Collect Feedback (route a document or item for feedback), Collect Signatures (route a document, workbook, or form for digital signatures), Three-State (track an issue, project, or task through three states or phases) and Publishing Approval (automate content routing for review and approval) workflows.

But sometimes such workflows doesn’t fit exactly to your scenario or worse, the workflow usage is just an overkill.

Next, I’ll introduce you a clean state machine (also personal) approach to handling state transitions of SharePoint list items without workflows.

Event receivers as workflows alternative

As you should suppose the only way implement this is by using event receivers (a.k.a. event handlers) instead workflows. But how make a clean solution event handler based to implement a state machine. 

Allow me show you how the final code looks like for a customized publication process based on this approach:

[StateMachine("State", typeof(PublicationRequestStateMachineValidator))]
public sealed class PublicationRequestStateMachineItemEventReceiver : StateMachineItemEventReceiverBase
{
 [State("Approved")]
 private void OnApproved(SPItemEventProperties properties)
 {
  /*...*/
 }

 [State("Rejected")]
 private void OnRejected(SPItemEventProperties properties)
 {
  /*...*/ 
 }

 [State("ReadyToBePublished")]
 private void OnReadyToBePublished(SPItemEventProperties properties)
 {
  /*...*/
 }

 [State("ReadyToBeUnpublished")]
 private void OnReadyToBeUnpublished(SPItemEventProperties properties)
 {
  /*...*/ 
 }
}

Notice the introduction of a few new classes:

• StateMachineAttribute: To indicate the column to be monitored and its validator class. The example above indicates that the column name is "State" and the validator class  is typeof(PublicationRequestStateMachineValidator). 

• StateAttribute: To indicate the event method that will be called when the state change to the specified value. For instance, the usage of [State("Approved")] means that the method OnApproved will be called when the item change to  the "Approved" state.

• StateMachineItemEventReceiverBase: Implements the base behavior of the state machine (invokes the validation to avoid not allowed transitions also call the event methods).

To make this work properly you must register you state machine item event receiver on the list that you want to monitor for state change. This must be done via RegisterEventReceiverIfRequired extension method for SPList. This method not only registers the event receiver the list, it also adds a custom column to the list to implement the change detection approach whether the event receiver inherits from StateMachineItemEventReceiverBase. 

This could be done with the follow code in a feature activation for instance.

publicationRequestList.RegisterEventReceiverIfRequired(SPEventReceiverType.ItemUpdating, typeof(PublicationRequestStateMachineItemEventReceiver).Assembly.FullName, typeof(PublicationRequestStateMachineItemEventReceiver).FullName);
publicationRequestList.RegisterEventReceiverIfRequired(SPEventReceiverType.ItemUpdated, typeof(PublicationRequestStateMachineItemEventReceiver).Assembly.FullName, typeof(PublicationRequestStateMachineItemEventReceiver).FullName);

How validate state transitions?

Support transition validation is also a cool feature of this library. In order to validate the transition, you can inherit from StateMachineValidator class and type your own. Our custom publication request example the state machine validator looks like this:

public sealed class PublicationRequestStateMachineValidator : StateMachineValidator<string>
{
        public PublicationRequestStateMachineValidator()
        {
            this.AddAllowedTransition("WaitingForApproval", "Approved");
            this.AddAllowedTransition("WaitingForApproval", "Rejected");
            this.AddAllowedTransition("Approved", "ReadyToBePublished");
            this.AddAllowedTransition("ReadyToBePublished", "Published");
            this.AddAllowedTransition("Published", "ReadyToBeUnpublished");
            this.AddAllowedTransition("ReadyToBeUnpublished", "Unpublished");
        }
}

Now if someone tries to change the state from WaitingForApproval to Published - for instance - the change will be reverted automatically.

This is also useful to enable or disable some actions. Here are couple of pictures that depicts the enable state of the ribbon button depends on an allowed transition validation implemented in javascript in combination with a REST service.

Example A: Enabled because transitions are allowed

Example B: Disabled because transitions are not allowed

Conclusions

This post is an introduction to the new StateMachine.SharePoint library. This library allows you simplify the implementation of a state machine based approach to monitor and validate the states of SharePoint list items.

For now, you can build this library from its sources and deploy directly into your SharePoint farm or wait for the "top secret" weapon and forthcoming project PackageManager.SharePoint  ;)

Can NDepend 6 and SONAR work together?

Introduction

As I wrote in this post, one of the greatest feature of NDepend were its “great Visual Studio integration in order to display your technical debt directly inside the IDE”.

NDepend can also be integrated as part of your continuous integration pipeline in order to make the analysis of technical debt results public for the whole team (or just break the build under certain conditions). There are a lot of official documentation about how integrate NDepend 6 with Team City (as a build server) or SONAR (as a quality metric tool).

But let’s start with my own experience setting up NDepend 6 and SONAR and checks the benefits of integrates both tools.

Integrating NDepend with SONAR

NDepend 6 comes with support for SONAR integration. The process is pretty forward and is well described on the documentation. After follow such steps you will get all NDepend’s rules imported into SONAR and you can activate them into a Quality Profile for instance the Full Analysis for C#. 

NDepend rules imported in SONAR

The issues could start when you run an analysis for large projects with several violations. If you run (the runner) with –X the stack trace will show you the java.lang.OutOfMemoryError as the exception. But nothing that can’t be solved following this recommendations: 

1. Run sonar-runner with x64 JRE. 
2. Increase the heap size by turn this line:

%JAVA_EXEC% %SONAR_RUNNER_OPTS% -cp "%SONAR_RUNNER_HOME%\lib\sonar-runner-dist-2.4.jar" "-Drunner.home=%SONAR_RUNNER_HOME%" "-Dproject.home=%PROJECT_HOME%" org.sonar.runner.Main %*

             into this one

%JAVA_EXEC% -Xmx3062m -XX:MaxPermSize=512m -XX:ReservedCodeCacheSize=128m %SONAR_RUNNER_OPTS% -cp "%SONAR_RUNNER_HOME%\lib\sonar-runner-dist-2.4.jar" "-Drunner.home=%SONAR_RUNNER_HOME%" "-Dproject.home=%PROJECT_HOME%" org.sonar.runner.Main %*

            in the sonnar-runner.bat file.

The results

SONAR is an isolated server that receive the results from “inspection agents”. Actually the inspection results are committed directly into SONAR database and the SONAR web application or dashboard displays the results in a centralized way. 

So, as you expected (and so do I), after run an inspection via sonar-runner, the NDepend's rules violations are displayed as SONAR's issues, just like this.

NDepend's violations as SONAR's issues

So, now you can manage (assign, resolve, or comment) such issues trougth SONAR interface. 

Conclusions

As you should know at this point, the answers is yes. NDepend 6 can work together with SONAR. But as you can also see I have a lot of work to do. So, I’m not sure what I’m doing writing this blog post ;).

SONAR dashboard

PS: This is not a "Do as I say, not as I do" post. As I also said, the "important thing is, not to accumulate technical debt and fix it as soon as it is detected". The sample reports shown on this post intentionally includes source with a tons of defects. Most of them comes from tests / PoC assemblies and auto-generated code.

Self-disciplined Agile Monitoring

Note for readers: I wrote this post over a year ago (even more), but for some reason I forgot to post it, so here it is.

Introduction

Now days, we are improving our development process laying-out our organization strategies, development process and methodologies. 

Our experiences point to a mixed approach, with the self-organization from Scrum and self-discipline from eXtreme Programming (XP). But about this kind of “mixed martial arts” for software development approach I will talk in forthcoming blog post. 

As you can read in one of my previous blog post, we use Team Foundation Server (TFS) as issue tracker and thanks to Scrum for Team System v3 (STSv3) process template, we have an “excellent” implemented guidance to execute Scrum “as is”.

But the existing day by day sprint monitoring’s tools around this process template hides the real behavior of the team in the iteration. 

On the other hand, we also have some needs about monitoring. Indeed against the theoretical practice of monitor the whole team we have to track individuals.

Some projects leads (me included) want to track theirs developers work against the importance of monitor the whole team behavior. Sam Guckenheimer (with Juan J. Perez), in his book “Software Engineering with Microsoft Visual Studio Team System”, aims us to use descriptive metrics rather than prescriptive. But here such measurement method “doesn’t work”, we need more control and also need a single view. 

Tracking the daily work in STSv3

Scrum for Team System (v3) is a great process template. Its major advantage of this implementation for TFS is about the usage of the server side event notification API. This allows updating all of computable fields for instance: start and end time for sprints (summarized from team start and end sprint dates), remaining hours for sprints and team sprints, and so on.

Scrum for Team System also comes with a lot of reports, but about daily work sprint monitoring comes with only one (maybe two).  This report is known as Sprint Burn Down. 

The burn down metric

The Sprint Burn Down is good metric but it’s incomplete. Such single line indeed hides the real behavior of iteration. 

Please, try to answer these questions:

• How can you notice if some tasks where moved out of the current sprint?
• How can you notice if some user stories where moved out of the current sprint?
• How can you notice if some tasks where added to the current sprint?
• How can you notice if some user stories where added to the current sprint?

The fact is that as this chart doesn’t display the planned work so it hides these behaviors. 

Notice:

1. If you see an ideal chart, an straight down line from an amount of hours as remaining work (at some point of the sprint) to zero (at the end of the sprint), doesn’t mean that everything is fine. May some tasks were moved of the sprint.
2. If you see a horizontal line doesn’t mean that everything is wrong. May some tasks were added at the same time that others were actually done. 

The burn down metric

The fact is that you are not able to answer this question: What had really happened here?

Complementary tools to track the daily work

One of my favorites tools to track iterations, and also share the iteration status with the whole team is ScrumSprintMonitor. Yes, the multi-process template screen saver. 

The effect of the usage of this tool in the team’s focus is incredible and of course, more over the “guys in red” ;-).

Demo screenshot of ScrumSprintMonitor from codeplex and ScrumSprintMonitor in action but in planning mode in a lab

I like the Scrum Sprint Monitor and have been using it for years, even when I had no longer available the TV in the picture above. But again, the main metric is the Burn Down.

This tool includes a lot of info so I get inspired and wrote my own, just like I thought that it should be written ;-).

Self disciplined agile monitoring

Well, the thing actually started some years ago when I found this chart in the book “The Art of Agile Programming” from James Shore and Shane Warden. Its name is Burn Up. 

Basically consist of a couple of lines. One to show the total of planned work and the second to display the progress.

The burn up metric

In the context of a project to help my own organization to get the right way in terms of software development practices, I bought some time in order to port this metric into this monitoring application.

My own ScrumSprintMonitor in action with the burn up metric

As I said before it was inspired in Scrum Sprint Monitor but it was re-written from scratch to focus in the Burn Up metric. The current version it’s only compatible with STSv3.

Conclusion

1. Now we have a very cool monitoring application with a very meaningful metric. The Burn Up. 
2. In order to distribute as fast as possible this tool and its updates across my organization, I also implemented a draft of NuGet based automatic update system. After done, someone told me about something called Shimmer. I have to review it ;-).
3. I almost forget. This application is powered by Catel and Prism in combination (a.k.a. Catel.Extensions.Prism). 

Happy sprinting and monitoring ;-)

Why should you start using NDepend?

Introduction

This weekend I finished painting my apartment. Then I laid on the floor, looking at the ceiling, started taking some shots and talking to myself:

A) Yes, it looks great and with this new lamp in the middle of the living room, everything looks perfect.	B) But what is that I see at the corner? Let me get a closer shot.	C) Oops!!! I made a mistake, I need to fix it ASAP ;)

Yes, I know, I’m not a professional painter. Therefore, I don't have the right tools to alert me about these technical debts while I’m painting. 

Such experience reminded me that I’m a software developer and recently acquired the right tool to detect such “dark spots” – a.k.a technical debt – of the source code while I’m coding. Its name is NDepend.

What is NDepend?

SONAR Web Report

NDepend is a static analysis tool for .NET managed code. As you should know static analysis is about analyzing code without executing it and is generally used to ensure conformance with the coding guidelines.

NDepend is not the only tool available for static analysis code for .NET, there are several tools including Code Violation Detection Tools like Fxcop, Clocksharp, Mono.Gendarme or CodeIt.Right,  Quality Metric Tools like Nitriq, SONAR or NDepend itself, or just Checking Style Tools like StyleCop, Agent Smith.

Actually, I currently use SONAR with its seamless integration with the build process in order to continuously manage code quality in centralized reports of technical debts.

NDepend also has integration with the build process, but from my point of view, one of its key features is the great Visual Studio integration in order to display your technical debt directly inside the IDE.

NDepend Dashboard in Visual Studio

Let's take a look at a very quick start with NDepend.

A very quick start with NDepend

Quick access
to the violation results

After installing a plugin and setting up your project, you should run a code analysis just by clicking the option from the menu NDEPEND => Analyze => Run Analysis or moving de mouse over a circle in the notification bar and click in Run a First Analysis on this NDepend Project.

If you move the mouse again - once the analysis finished - over the circle in the notification bar you should see the Code Queries and Rules Summary, with fast access to the Critical and Rules Violated. 

Such results, categorized into Code Quality, Object Oriented Design, Design, Architecture, Layering, Dead Code, and so on, are displayed on the Query and Rules Explorer panel and allow us to navigate from the violation directly to the source.

Categorized violation query results on the Query and Rules Explorer

For instance, let me check from Code Quality category, the rule Methods with too many parameters - critical.  

The rule description is the following: Methods with more than 8 parameters might be painful to call and might degrade performance. You should prefer using additional properties/fields to the declaring type to handle numerous states. Another alternative is to provide a class or structure dedicated to handle arguments passing.

The analysis found 16 violations of this rule, by clicking it you can navigate to the method. In this case, I selected one with 9 parameters and found out that indeed, it must be refactored. Now, thanks to the integration of Visual Studio and Git,  you can also see who's the author of this violation. 

Navigating from rule 'Methods with too many parameters - critical' result to StartSiteCreationProcess method

Let me take a closer shoot to see who that is:

Oops!!! it's me, I need to fix this ASAP ;)

Conclusion

The important thing is, not to accumulate technical debt and fix it as soon as it is detected. For .NET developers, NDepend is the right tool to start with.

You can make mistakes (critical or not), but being aware of your code quality constantly makes the difference between the apparent and intrinsic quality of your sources and consequently your products; even when untrained eyes may only see the beautiful lamp.

Btw, It seems like I've got similar skills as a painter than as a software developer ;)