Collected works and current thoughts
I started working on a book some time back on C++ and something like OOD or TDD or some such theme. You can see its current state here. I did that because at the time I was teaching C++ to C programmers at NASA and I wanted to have additional background material for the students.
As you can see by the date, I haven’t done anything with it lately. I did contact some publishers but nobody seemed very interested and, like with a lot of my side projects, I get bored easily (my biggest known weakness), and have not been doing anything with it.
Then at the end of June 2012 I get the following email (reproduced with permission but name left anonymous by request - note, not a native English speaker but clearly able to communicate nonetheless):
I just read your WiP book “C++ and Object Oriented Design, The Least You Need to Know” and just really wanted to say how glad I am I found it.
I am lead programmer in small video game company (few years of experience in C++) and just recently started to look into TDD. I was really surprised that (as you also mention in your book) there are virtually no serious resources on TDD in C++ and as much as C#/Java examples can be helpful, C++ has its own set of challenges (as usual;)) in that area.
Main reason I got interested in knowing TDD better was the fact that I got more and more afraid to do changes in the system, so I really wanted some kind of safety net from unit (regression) testing. But thanks to your book I already made bunch of changes in the system I would never thought of being helpful. I learned that what “Driven” really means and I never thought I will change my way of designing C++ code so much after 3 days of reading. I can honesty say that last book that influenced me that much was “Modern C++ Design” by Alexandrescu few years back. I consider myself skilled C++ programmer, I use STL and BOOST daily. Metaprogramming is also not a black magic to me when necessary. Thing is, even tho I could probably describe most patterns and its usage from GoF book I must admit that after reading your book I am not sure I really ever got true OOP. It always felt kind of too Java for me and not really suited for “Modern C++”. Again, after your book I feel very different, relieved even, It shows that C++ code can be both modern and maintainable (I feel sorry for some of BOOST maintainers).
To summarize: please, please finish it. Not only I must know the rest of chapter 7 but outline of ch. 8 is like the worst cliffhanger ever! I feel that book is really necessary. I know I would pay a lot to have this on my shelve. And every next tome of it ;)
With new C++ standard there is so much more to discuss. For simplest example: which of smart pointers would you use as a factory method return type (in years I went from auto_ptr -> shared_ptr -> unique_ptr -> back to raw ptr because of all the limitations that standard ones impose)? How would move semantics affect your class design?
This is probably the nicest feedback I have ever received and now I’m going to get back into writing the book. I’ve changed the title from “C++ and OOD” to “C++ and TDD” and I’m trying to get it ported to a better format for e-book distribution.
We had a little bit of an email conversation and he followed up with:
sorry for the late response, I am swamped lately due to closing release of the game I am coding and you Sir are are the reason for quite a lot of this work actually as I was quite happy with the shape of my codebase until I read your book :P
Seriously though, I am happy to do this because with every smallest possible change I make, with every smallest test I add to the current code I feel better (more confident) about the system. I just stared with TDD game but I don’t think there is a return for me at this point. I know that tests can’t prove correctness of my production code but for me it is enough to know that it at least does what I expect it to do, even if that expectation is flawed I like to know that it is implemented correctly.</br/>
I am very happy to hear that you want to continue to work on your book! C++ programmers need it. I need it ;) Why (imho) are Sutter’s and Mayer’s books so popular? Because with C++ we already have way too much to worry about everyday. There is so many ways to do things “correctly” that it is often very hard for inexperienced (as myself) programmer to pick one. That is why we need guidelines from the better ones. Not a generic design pattern example (poorly) ported from Java but a real C++ code. Heck, I am now even changing implementation of my global registry of types (for Object class derivatives) because of your book (based on your MathOperation registry of course). It was actually very similar in basic concepts (macro for registration + global map of typeid -> factory functor) but your testable implementation is just so much better that I just could not sleep with mine obvious crappy one. Btw I realize it is just about the concept (client unknowingly uses default global registry) but current implementation of global RegistrationMap is not thread safe in C++98 (fixed in C++11).
Here’s what this says to me: He has groked it. Just a few points that say that to me:
- “I know that tests can’t prove…” - this just speaks volumes of pragmatism to me.
- “I don’t think there is a return for me…” - it’s the same for me, once I became test infected, writing automated tests was non-negotiable. TDD and test infected are not the same thing but I feel almost as strongly for TDD as I do for test infected.
- “I just could not sleep with mine…” - I’ve been there. In fact, last night I was reading a little bit planning to go to bed and ended up on the long side of 3 a.m 2+ hours later.
I want to give specific credit where it is due. The “MathOperation registry” he speaks of is based heavily on CppUTest. I copied what they did, figured out how to design it to make it testable, then how to build it up in an approachable order test-first (as opposed to test-driven), but the basic design is not mine.
In any case, here I am getting back into some side work on C++ because of this person. I had clearly let the work slide. Now I just have to finish a blog entry as a guest blogger for Martin Fowler’s blog and port all of that work!
