Organizations reorg all the time. And again. Why do they do that? Setting cynicism aside, organizations reorg to adapt to new realities, to new demands. A team of 5 people that grew to 20 people needs to split to smaller teams. A business group dealing with a fast growing market needs to come up with a new strategy to cope with the demand. A startup of 20 people will need a different structure than that of a company of 100 people. As business demands change there is a need to adapt the organization’s structure.
Reorg is an expensive venture, yet organizations do it again and again. Because they have to do it – they have no choice.
In a similar manner, the codebase of a product needs to be reorganized again and again to adapt to changing circumstances. A class that has more and more methods needs to split to smaller classes, otherwise it will be very difficult to maintain it. A conditional (If-Else statements) that grew up to a monster needs to be shrunk again, otherwise it will be prone to defects. A once simple interface that grew and expanded needs to have some wrapper to make clients’ work bearable.
In fact, taking a hike on Conway’s Law it makes sense that refactoring will somehow be related to reorg, at least in the sense that a change in the organization structure is related to adaptations in the codebase.
We call the process of adapting the codebase “refactoring” – changing the structure of the code not to get new functionality but to make the code better adapted to our new demands, business and technical alike.
Unlike reorgs, refactoring is part of the ongoing work of every developer. Every time a developer is handling code she needs to think whether she should change the structure of this code. Should the method be renamed? Should a new interface be extracted? Should she replace the conditional with subclasses?
Just like dev doesn’t work without test, dev and test don’t work without refactoring. There are many cases where refactoring is a key player for making the code testable (but I’ll write about it some other time).
I’ve never been in a scouts camp but I hear they’re supposed to leave it in better shape than they found it. In the same spirit, a developer should leave the code in a cleaner state than she found it.
Two issues immediately arise, though: First, refactoring takes time. Correct. Reorgs also take time, yet we do them. Refactoring takes less time and provides faster results.
The second issue is that constantly refactoring will make the system change all the time. Changing names, changing structure. Won’t that be counter productive? Won’t that inhibit maintenance? “I’m used to look for method err4get and now someone renamed it”. The idea is that refactoring should make the system more maintainable. If it makes it less maintainable – don’t do it. Names should be clearer, structure should be easier to understand, easier to test and change. Getting into a state of mind that we’re not afraid to make changes in our code is a healthy thing.
Every time your organization goes through another reorg you should ask yourself when did you invest such efforts in your codebase. Codebases reorg should happen all the time, on small scale, getting the code ready for the coming business challenges.