Software Production Management

Eight+ months ago I decided to bring a change to my life. I decided to resign to my Account Director position at Studiocom. I was sure that even when Studiocom was named as one of the best place to work here in Colombia, there was no opportunity for me to grow personal or professional. I was working hard more than a year on amazing projects, getting results for the company, and the client, but with no professional opportunities, or any recognition or reward, feeling frustration and of course, leading me to a no-return situation.

Then, I took a chance with a well-known entrepreneur at Torrenegra Labs – thanks to Leo Suarez and Alex Torrenegra. It was not a smooth transition. At the beginning I was supposed to work on a project (shh! It is a secret one), however after our first formal review on the scope and budget, we decided to keep this initiative on hold. It was back on December! OMG! What am I going to do?

Fortunately for me, Alex invited me to join the LetMeGo Team as the owner of the Referral Program (during my first three months here at Torrenegra Labs I offered my help on LetMeGo project giving a hand to launch the Beta on time). I joined the team last January. So, “What? Referral what?” I was confused and I was feeling ignorant, what is a referral program? and how am I suppose build a successful referral program?

I started working as Architect of the LetMeGo Subsystem on charge of the Referral Program, and then of course, as any good Architect, I started coding, but not coding like I was used to. That is when RRapido Methodology and I formally met. Even when it is a mix of Agile, Traditional Development and Test Driven development, I have to say that the most valuable asset that we have is our methodology. I’m not saying it is perfect or a “One-Size-Fits-All”, however I will give my impressions on it:

• Applied Test Driven Development reduces the chances of getting errors on late development phases or iterations (I will not discuss here if they are the same or not).
• Extremely well document UC, TC and UI – LMG is the first project I had participated on were tech docs are almost perfect and detailed good enough to allow clear trace between code, tests and UC – it is a little bit confusing because Agile Manifesto includes a line “Working software over comprehensive documentation” – what does it means anyway? Even on traditional approach you are always looking for “working software”.
• Noisy communications – RRapido fails on how things are communicated. We had what we call here in Colombia “meetin-gitis”, but through Skype. Most of the times we over communicate everything to everybogy. With time you get used to it, but it is like be in a party with everybody talking each other and with loud music.
• Microblogging looks like useless. We use Yammer, I microblogging tool for private(?) teams. We used it to promote our tasks every day -I’ve to admit that I’m not most dedicated team member to that.
• Small group with one “I-know-everything-about-the-system” manager. RRapido, and the most significant and valuable addition to our team is what Scrum calls the Product Owner. Alex is our product owner, he is a technical person that has developed a great “business” sense. With his experience and dedication RRapido and the whole project is kept on track. There is no doubt about it. One day without Alex is a day when things slow down (a little bit)

So it is time to talk about Mr. Alex or  Don Alex. I disagree about few practices on Alex management style however, and based on the facts and the results I say: He is on care of EVERY detail of the system, leading the team to a successful product and “working software”. He is the dream of every development team, a dedicated and objective Product Owner that understands technical challenges and risks of every decision. Alex does not ask for outstanding deliverables all the time, but when he does that, he also provides the tools and time to succeed.

RRapido Methodology depends 100% on the Product Owner -what we call Product Developer. He/she is the key of the RRapido Methodology, as the keeper of the process. It is so difficult to know about the system, the business, the market, the customer drivers, to have relevant technical background on the development, to have passion about the business, and also to be the process owner, that RRapido becomes a process for few -maybe a process that can be adapted for those who had worked with Alex.

Recomendations

As my dad always says, if your are complaining about something that you can change, why you do not change it. I decided then to write this post, sure that Torrenegra Team will read it and probably -if I succeed- will include few process to the methodology.

• Methodology is a set of steps, it is not a ToDo list. RRapido feels like a set of things to do, but I would prefer if we can set them as a procedure.
• Communications are noisy, One-on-One meetings are great, Publishing Notification Board is ok, all other are noisy. I would prefer a message board or workspace site.
• Microblogging is useless when weekly tasks have been set. Remove it or use it to assign tasks, not to report them.
• Documentation. I’m in love with documentation, it is not perfect, but it supplies all developers need to code, test and validate the required functionality, however, navigation between UC, UT and UI is painfull. I would look for another way to link UC and UI (maybe a two column page or something similar).
• By formalizing RRapido steps, dependency on the Product Developer role will be reduced.
• Test Driven Development Rocks!

To choose the right software is not a simple process, and to choose the right project management software is then an even more complicated process and decision. Below you will find few tips and proposed procedure to reduce the risk inherit to this decision.

Tip #1

To choose and implement a PM software tool is not the same of implementing a PM process. Many organizations try to implement a software tool expecting a nonexistent-unnatural process improvement. If you do not have a formal process, or even if your process is not working you have to stop thinking that automation will fix/improve/solve your problems. Automation should be used to reinforce a process and minimize the weaknesses.

Tip #2

Every software implementation process includes at least the following steps -of course every company has its own natural process. Public companies have also additional restriction and evaluation/validation processes that will extend the suggested process:

• Identify needs
• Define selection criteria
• Create a list of options
• Create the request for information
• Evaluate responses
• Reduce the list of options
• Ask for demonstrations or pilot programs
• Choose one

Tip #3

Try to isolate your needs by using the following dimensions to measure the products you are considering:

• Scale – how big the change will be?
  • Simple: are you going to organize your projects?
  • Personal: are you going to automate estimation and planning on you projects
  • Collaborative: are you going to support a team? are you going to share information? do you need to centralize team communications?
  • Enterprise: are you affecting the whole company? are you going to bill to your clients using your projects’ data? do you have virtual teams all around the globe?
• Management Paradigm – do you and your team follow a traditional or agile approach to project management?
• Process Maturity – how formal/strong is your process?
  • Chaotic: No evidence of documented processes or best practices
  • Active: Documented processes carried out, but not formalized
  • Efficient: Consistent discipline started
  • Responsive: Ubiquitous and measured
  • Business driven: Provides data and information to drive business decisions
• Implementation model – are you going to buy the product and support it by yourself? are you going to adopt the SaaS model?
• Budget -
  

Tip #4

Set your goals – do not expect to do everything better and to include any improvement during the first phase (or the initial implementation cycle). Prioritize to get faster results. Below you will find a list of possible goals that you could address with a PM software tool

• Improve project reporting and tracking
• Improve estimating and scheduling
• Reduce cost or speed process up by automating workflows
• Improve resource assignments
• Improve project communication
• Improve project team collaboration
• Improve overall project process

Every goal will impact different functional areas within an organization. You should plan your implementation to impact those areas and improve those process that will add the most value.

References:

• 7 Goals of PM Software @ Software Projects.org
  
• Software Selection @ SoftwareProjects.org
• Evolving the Maturity Level of Your Project Management Office @ CA

Few years a go, Ruby on Rails (RoR) start shinning because it was only framework able to speed up the Web development with a set of all-in-one tools. As Java experienced programmer and architect, I’ve to recognize that the RoR team made their point. I still remember that video with a guy with a lot of jars on top of a table (check it here) that made me LoL for a while -an awful truth. Now we have a lot of solutions that offers similar out-of-the-box scripts and frameworks that build code as RoR does -the fancy but power/useful scaffolding, the code generators, and others. Few of them:

• Seam Framework -great tool
• Django -for Python developers.
• CakePHP
• Code Igniter
• Symfony
• Zend Framework

The list could be huge, but the main point is not to check them all but to provide few tips to start a Web-based project and choose the right tool.

The .NET and Java mistakes
I do love Java, I’m a fan-boy of JEE technology and portability but, I’ve to say, even if it hurts, Java is a huge rock. It means you need to have a huge infrastructure, a huge processing capacity and a lot of tools to start a JEE project. Why? Let’s face it, you need more than your imagination to start a project and it isn’t easy nor cheap to find a great hosting solution, to properly setup an environment for development (with sandbox and preproduction/test environments). Even if your project is really small it is a good idea not to have the test environment on the developer computer -I hope you don’t think this is a joke, it happens ALL the time.

So, instead of being easy to start with a basis, you have to deal with complex configuration problems -it is good to have all those options if you are on a banking/financial project with a couple millions as budget, but in a real economic-crisis world you should look for “point-n-shoot” solutions -if you know what I meant.

The Apache Friends
So you have a lot of alternatives on the Apache HTTPD side, with PHP (the popular), Perl (the bad cousin), Python (the genius), RoR (the cocky brother) and all others.

I’m not a scripter guy, the compiling hell is for me but, the simplicity shows us that you don’t have to deal with complexity all the time for Web projects. So, give non-corporate-supported technologies a chance. It was hard to me, but they worked faster and cheaper than expected -there is not measure to say better. It is cheaper to create a top-class development and testing environment and work with remote teams if you use open solutions -not necessarily open source. It is not all about the language you love, the server you own, the database that rules the world. It is about cost, simplicity and efficiency. It is easier to get/hire great developers on these tools than for Java – .NET is for chickens and everybody knows it (check this article if you do not believe me). Software development is not about coding, is about solving, that’s what Microsoft doesn’t understand. They code a lot but they do not solve too much, right? You should notice since Windows 95 and the coming of the Blue Screen ERA!

The Automagic
Before I became a senior developer I worked migrating some applications from Oracle Forms and PowerBuilder to J2EE -we used to put a horrible 2 in the middle few years ago. And those tools were great for the task they were created. So it took almost a decade to have those tools on the Web -and they do not work as fine as those mentioned. However, the Automagic isn’t new at all. ActiveRecord exists since, I don’t know, the Z81 maybe? He He He!

So now we have tools that reverse the database and create ORM models, pre-formatted pages and objects that speed up -years- the development. So you can concentrate in other IMPORTANT STUFF like, User Experience (UX), Information Architecture (IA) and the forgotten design.

The generated code is clean as possible (I remember the JBuilder and its generated & encrypted code). And this is thanks to the separation of concerns and excellent pattern implementations – the commonly used for the Web is the MVC supported with other ones like ActiveRecord and IoC, but you can find a whole universe on each platform and framework – I do have an old poster of J2EE patterns as wallpaper for example!

The Verdict
Before fan-boys appears into the scene while you are deciding what platform should use, please take a look to the ones you probably will not look at -don’t waste your time with .NET . During these 7 (should I said 8?) years working on Web development I’ve noticed that only big companies uses robust (slow development) platforms like Java and .NET -by big  companies I meant with more than couple billions of dollars on their budgets. They spent a lot of money in servers and infrastructure and ironically most of the time they use always versions old versions of platforms, servers, databases and all. So even if they spent a lot of money they cannot move as fast as it happens with the technology. Microsoft still produces Adobe Flash banners right, even if they have their own Silverlight. And it also happens with JavaFX. When you develop Flash components some companies still requests Flash 6 or 7, when we all know that we have version 9 and 10 running on most of the computers on the web. But hey, they want to be sure and even those computers that are off since 1995 can see them! – why?

Before you spent more then 6 months working on a project without a single release -could be a partial one, but functional- keep in mind that probably it will take other 6 months before you see the product/result you are expecting. Automagic means fast, so I hope you do not spend more time thinking on if you should or not, and how, than coding… Otherwise it will demostrate my point

More than ten years I was forced to wait to see a real web based project management tool -by real I mean a decent software product that seems to address the PM needs and it is not a poor desktop application pushed to the web -like Microsoft Problem Server -oops I mean Project Server. Browsing the web I found the PMXPO 2008 and the SIIA CODiE Awards 2008 pages. Looking at the CODiE Awards 2008 finalists’ list I found @task, a company dedicated to deliver real software that uses the web.

I will try to get a demo or at least a try to that software product. As developer I know a lot of things about web development but also as PM I feel frustrated about the poor quality of the products delivered -that is one my personal reasons because I decided to start the SpZ Framework project. However, there are hope on new products that are jumping in and doing a great work.

PM should not have to deal with IT project strategy however almost all the time PM are “abandoned” in the middle of nowhere with a simple request: build a software tool that probably the client do not understand at 100% and it for sure is not documented.

The first steps are to (re)build the project scope and have a plan, but in the meanwhile your new team will need to have the tools to start their work. As PM you probably will have your own tools and templates but the team will have an IDE and that’s it. You should not deal with technical issues, but the reality is that as PM team will depend on you to get the right/optimal resources to simplify the developers work.

If you are those lucky guys that receive a software project since the beginning and not sometime after then you will have a better chance to get the right resources. My recommendation: Keep it simple but useful -even notepad is a useful tool if you use it correctly.

Your team needs:

1. Deal with versions. Even before start your coding phase you MUST have a concurrent version system. In the market you will find a lot, but probably the common ones -and open source are: CVS and Subversion (a.k.a. SVN). Versions will force the team to work closer and with the same structure. If your IT team do it well, you will appreciate it a lot.
2. Common IDE. Developers have their own affairs with specific IDEs. But you are the PM, ask your IT lead and force them all to use the same. I don’t like democracy, PM rules! However, as developer, I love Eclipse IDE.
3. Documentation System. Most of the developers around the world will hate to document their code or applications. At least here in Colombia, when you ask developers to document their code or provide formal documentation they will give you the bad-eye look (in Spanish the malde’ojo) and you will understand that not the greatest documentation will be available at the end of the project. So provide the team with the tools to make this pain softer. I recommend only WYSIWYG tools like a Wiki, Google Docs, or any other fancy tool.
4. Track their progress/work. Development tracking tools are suspiciously linked to Bug Tracking but it is not necessarily the truth. I’ve been using bug tracking tools like Mantis BT and Jira as Tasks Assignation tools. Keep in mind that a Bug is a request to do (fix) something. So why you cannot split your work packages in work items and assign them to the team -as you do with bugs.
5. Provide test environments. I have to say this, why non-developer centric companies like Creative Agencies building Web applications do not understand that a Test Environment IS NOT the developer’s computer. What is this all about? What is the mystery? Test environments are needed, mandatory needed. Release building is a complex task that will have a lot of manually interactions -team pushing changes and enhancements- and it is NEEDED to have a place where to put all together and validate it before release it to a client -even to a client test environment. Please do not cut costs here. I will start a facebook group named “Why we do not have a test environment?” -BTW I do not like facebook, I do not even use it.
6. Put it all together and working. This is the real thing about IT support. Install software is a piece of cake, is not a complex task, put all together and make it work smoothly is the key. So ask you IT team to do it, or do it yourself, like I did for some of my dev projects. There are many ways to integrate everything but I’m giving you some useful links to found the way to do it:

• SVN, MediaWiki and BugZilla working together
• MantisBT & DokuWiki
• MantisBT & Subversion

SpZ Team is currently designing an All-in-One integrated solution. For latest updates check the Framework page.