Iterative Software Development Cont.....

My last post gave an introduction to what iterative development is all about and what kind productivity and value addition it brings to the table. In this post i want to address the questions unanswered in my last post. Cant keep the readers in suspense now can i ;) ... But hey everybody likes a little thriller every now and them right? :) .... So what were the questioned that were unanswered in my last post? Ok ok if your bored to go back to it and check it out let me list it down here again :D....

• How to approach iterative development
• How to project your iteration size
• How to incorporate new customer changes half way through an iteration

Taking first thing first, the way you should ideally approach an iterative development process is by first analysing the initial customer requirements you gather and breaking down those features with respect to the application scope. Then what you should do is to weight every feature according to the priority level (High/Low) as perceived by the Customer and estimate the number of development days each feature will take to complete.

This process includes the asnwere to our second question which is how to project your iteration size. So you can see them both as interelated questions. Iteration size normally is recommended to be best kept at 20 working days(i.e one calendar month). But ofcourse this is definitely a varying factor depending on the complexity and the size of the project.

What you do after you decide on your iteration size is to try to calculate what features can be included in the first iteration which keeps in track with the iteration size you specify. But also you should keep in mind to put in most of the high priority tasks to each iteration as possible. Sometimes there maybe some low priority features that need to be provided before you can continue with a high prioratized feature in which case it is acceptable to include low priority features into an iteration.

Going on to my last question which is what you do if the customer comes up with a new set of features while you are in the middle of one of your iteration. Again you should go in the normal flow as explained before and weight new features accordingly and then try to see how it best fits the current iteration. If it is possible you could always push back some of the planned features of the current iteration to the next iteration and incoporate the new features whilst keeping the iteration size intact. Then you could ask the question what if its the last iteration and the customer comes up with a new set of features. Well in this instance you have to come into an agreement with the customer as to what is realisticaly possible to achieve for the given deadline and either try to move the dead line or add more developers or ask the current team to work long hours to make the current deadline. But in my perpective i do not believe adding more developers or making the existing ones work long hours will do the project any good as overworking burns out developers which in turn will lessen their productivity in the long run. And adding new developers in the last iteration is or for that matter in the middle of the project is a very decisive decision to make because according to what is stated in the book The Mythical Man Month by adding new developers you are adding n(n-1) / 2 communications channles within the project where "n" is the number of people in the development team.

Hence it is always best to try to negotiate with the customer on the current situation and try to shift the deadline to a latter date or to come to an agreement and compromise some features as deemed appropriate by the customer that can be included in later versions after the deadline.

Well thats about it for now about iterative development. Two pretty long posts on iterative development huh?... :) .. Guess you guys are now getting sick of reading about iterative development. So ill try to blend in something different in my future posts to come.

Until then its adios from my side and happy development (hopefully iteratively :D )