Coccinelle: 10 Years of Automated Evolution in the Linux Kernel

Tell us a little bit about the work that you're doing today.

Mostly I'm interested in the maintenance of very large codebases. I would like to make it easier for people to understand the code, to get a good feeling about the code, to change the code and especially to understand the history of the code and how they can use the history to help them go forward in the future.

What is the goal of the talk?

I want to make it easier for people to update their code in a good way. For example, some code may be written in a certain way. Maybe it was a good idea at the time, but one can see that it's not a good idea anymore, and one would like to update it. But then one may think to onesself, oh, I'm going to have to go through 500 files and change 600 lines of code or something like that. There's a very high chance of making a mistake. It would be also extremely boring. And so I'm not going to do it. I'm just going to live with the way things are and maybe things will be okay somehow in the future.

Can Coccinelle be used for other languages than C/Linux?

Today it's targeted mainly to C code. We've developed it with Linux kernel in mind, but it has been used for plenty of other C codebases. There's some issues with respect to macros and you have to configure it with the macros that are using your codebase, but it's usually pretty easy to get it to work pretty well on most C codebases as long as they are written in a reasonable manner. We also support some aspects of C++. If you want to update the part of your C++ code that looks like C, then it may work on that as well. And we also have a prototype for Java that hasn't been explored very much, but if people are interested in using it, then we would be happy to have feedback at least and to improve it if necessary.

What is the persona targeted by this talk?

Most of our users so far have been developers, that is the people actually touching the code and have some recurring change that they need to automate. Coccinelle can also be used for finding certain patterns of code, maybe patterns that might have bugs. And so a target would be people who need to find those kinds of issues. In coming to QCon, I hope also to be able to talk maybe a bit more to higher level managers who might be interested in supporting further development of Coccinelle or just using it maybe on a wider scale in the projects that they're working on.

And perhaps expanding it into the Java language.

We're also interested in expanding it to other languages, but we don't have the manpower for that at the moment. The Java port took about three months, so maybe other languages would not take too long either. I don't know.

What would be some takeaways from your talk?

The main point is the methodology of how you can use the tool. So in the talk, I will be going through that methodology, it's quite simple. You just take an extract of code that represents the thing that you want to change, and then you extract the code and make it a bit more abstract and then you just run with it. So I would like people to come away from the talk feeling that they can download the tool and they can know how to use it pretty much on their own. Maybe just based on the slides of the talk, they shouldn't have to spend a lot of time studying documentation and so on. It should be hopefully just ready to go immediately.