What do you do to understand some code that you didn't write? Or code that you wrote long time ago and don't remember what it does anymore.
Do you have some technique that you go about? Do you analyze the structures first, or the public methods, or do you draw flow charts, etc.? Or do you fire up the debugger and just step through it? Or do you just ad-hoc your way through until you understand it?
I use a mix of:
Not necessarily in this order :-) It's astonishing how much easier it becomes to understand what a piece of code does after some refactoring.
refactoring
the code, thus making it clearer, and in standard that ah-hock established.
Usually, I analyse a simple part first, e.g. the module used to maintain a small table. This teaches me the style the other programmer is using. If I have problems understanding even that, it's either very badly written or my knowledge of the language, framework etc. is insufficient. Once I grasp the simple part, it's time to move to the more complex parts of the program.
Try to fix a bug
Best possible way to get to know the code! :)
I go through the usecases. Each use-case starts at some point and finishes at another. Start looking at the begining and follow the flow. When you've examined three or four usecases you know the structure of the code.
Preferably you should be writing tests when following the code, since it will help you keep a more active role in examining the code than reading it line by line.
The debugger is a great tool to follow the flow, you could make some quick tests that doesn't realy assert anything but start the code at the point you want to debug from, to get a quick starting point for the debugger.
Though depending on how comfortable you are with tests, tests might be faster to check expected results.
this is coming with the experience. when you're a newbie, or you jump into other programming language is a little bit difficult. when you have several years working with a language, then is easier to understand.
but, as a general rule, I'm firing up the debugger, and start to understand what's happening in it. also it is VERY IMPORTANT to comment your code, and to work on documented code.
Use a white board to write out and diagram interactions between classes or methods. This can help you to see the flow of the program. Once you have the 100ft view, then start digging, tracing and debugging to find the nuances of the system.
Generate/draw/read a call graph.
How do I go about understanding others code?
Well, most I don't go about it at all. I only try to understand it if it doesn't work, and I'm trying to figure out if I did something wrong or "the other" did. And the tools I use for that is reading the code and using the debugger.
If that also doesn't help, I mail the author.
Different people have different learning styles, so you have to choose the method that works best for you.
The first thing I do (after building the project) is read the entire code base through at least once. That gives me a general idea of where everything is. Then I choose a section to examine in more detail. Data structures would be a good place to start. Once I have a general idea of what's going on, I do the same with another portion of the code that interacts with the first. After enough iterations, I have a good sense of how the code works.
Good question, I don't think I have ever sat down and documented the process.
I guess thinking about it now, I just read it:
Line by Line
Of course this doesn't always work, asking the author is usually the last resort.
I imagine how I'd write the code and look for similarities.
Interesting question, I don't think I've ever documented the process before but here are a few things I like to do.
assessing
doing grep -l [text to find] [files to look in] is sweet for this kind of thing in Mac or Linux. No idea how this would be accomplished in Windows.
researching
Taking the example that it's an internal project I've inherited--maybe from a long departed colleague who I've never met:
manual labour
The previous steps were all designed to prep you for the intense amount of manual labour that is inherently involved in reverse engineering someone else's code. Up to now, most of the work has been focused on reading.
Substantial refactoring should only be conducted if the state of the code system actually requires it (i.e. it's running really slow, it's become a "big ball of mud", etc.) What you want to avoid is simply trading in one bug for a new one. In my experience, refactoring a project just to wrap your mind around it has unforseen consequences, and these consequences may only become evident at a much later date. Even with automated unit tests, in a large system it can be difficult to re-test every single aspect of a development cycle.
When you're reading through a class, look at the attributes and behaviors that it has (methods + fields). For methods, take mental note of the method name, it's input parameters, and it's return type. If you don't know the method name (because it's obfuscated), then sometimes you can get this information by looking at the input, output, and implementation.
Take this partially obfuscated method as an example:
public double f(double d0, double d1, double d2)
{
double d3 = this.locX - d0;
double d4 = this.locY - d1;
double d5 = this.locZ - d2;
return MathHelper.sqrt(d3 * d3 + d4 * d4 + d5 * d5);
}
It accepts three doubles, returns a double, but the telling part is the mathematical equation: square-root of a-squared plus b-squared plus c-squared: That's distance!
Now, you'll want to refactor f() to distance(). Decypher the easy ones first. Once you have more and more of the puzzle solved, it'll give hints into the harder to decypher areas.
My favorite trick is control+F. That's the search function. And you can use to search entire projects too (not just individual files). How do you even know what to search for, you ask ? Well, by using the application or library in some way... it'll give you hints as to what to search for. First, you have a general idea of what particular feature you wanna look for. Then, if you've used the project even just a little bit, you can get more specific clues to narrow your search.
For example, I'm working on a project now that I've inherited from an inactive author (an open-source Minecraft plugin with 418 files and 46k lines of code). And we had a user open a ticket: "What is the permission node to use join signs ?" Heck, I have no idea. And the documentation didn't mention it either. No sweat, control+F to the rescue. (My weapon of choice).
One of the things you can't always 100% decypher is the question of their design: "Why did they design it like this ?" One can only speculate.
