Wednesday, October 31, 2007

How bad code is formed

Ignore in some case consultants do this purposefully to keep their contract, in general no one intentionally create unmaintainable code. However, bad code silently creep in over the period of the product's life cycle.

Manager's mind

Lets write the code quick. We don't have enough time for the next rollout. If we cannot rollout, the project may die quickly. Lets put long-term maintainability as a secondary priority, which we'll improve after the rollout when we have time. By the way, who knows if I am still working in this project after a year.
  • Unfortunately, if this rollout is successful, the next release cycle has an even more tighter schedule
  • Code is "written once" and "read several hundred times" throughout its lifecycle. Subsequent maintenance cost is several hundreds times more than the gain we made in its initial rollout.
I cannot justify any resources spent on refactoring because it doesn't add any features to my next release. The code is working already. You mention about improved code maintainability and I don't know how to quantify that.
  • This is how bad code retains and accumulates. After its momentum has grown to a certain point, then everyone just accept the reality that the code is already bad enough that no motivation is there to improve it.
Lets have the developers focusing in writing the real code. We'll have the QA folks worry about the testing. This way, they can proceed at full speed in parallel.

Developer's mind

Similar logic is working elsewhere but it doesn't fit exactly what I want. It isn't provide enough feature for my need and it also provide some features that I don't need. It is not worthwhile to spend my time to read and understand how it works, lets re-implement my version from scratch.
  • This is how duplicated logic getting into the system. There are many ways to do one thing.
Similar logic is already working elsewhere. It is not worthwhile to spend my time reinventing the wheel. Lets copy it here and modify for my need.
  • This is another way of how duplicated code is getting into the system. When you change the code in one place, you need to make corresponding changes in many other places where you have made copies, otherwise some copies may not be working correctly.
  • Due to "laziness", variable names are kept the same in their copies, making the copied code even harder to understand.
This code is not broken, please don't make any change
  • This is how bad code retains and accumulates. After its momentum has grown to a certain point, then everyone just accept the reality that the code is already bad enough that no motivation is there to improve it.

Of course the code is hard to read because it is solving a complex problem. If the code is easy to read, it must not be doing anything significant. You just need to be smarter when reading the code.

I am not convinced that this is the right way of doing this kind of things. Therefore, I am not going to follow the convention even it has been done in many other places. I am going to do it my way.

I don't know why I have to do it this way. But since this is the way we do this kind of things in many other places, I just follow that convention. Is "consistency" a problem ?

3 comments:

Anonymous said...

Good articles. Do write more!

Anonymous said...

Excellent!!!

Anonymous said...

"This code is not broken, please don't make any change"
- this is just what one manager in our company told me... sigh