Blog #2

Your blog entry must explicitly answer the following questions about the class (mark the questions in bold in your post):

• What are you most proud of doing last week?

Regarding the class, I don’t feel I did anything particularly worth being proud of. But I did finish and submit project 1, so let’s go with that.

• What ability or skill did you use to make that moment the most proud of?

I used many of the abilities I have learned over my career, but perhaps the most useful was the ability to quickly recognize when I don’t know something and seek help. I asked Dr. Downing for clarification on several of the imported issues where I wasn’t sure what was being requested.

• What impact did what you do have on other people?

I don’t think what I did will have much lasting effect. There were some responses to my comments on the “paper” (the syllabus), so perhaps I provided some amusement.

• What did you think of Paper #1. Syllabus.

I didn’t expect this choice for the first “paper”. I’d looked over the Papers page of the class web site, so I expected the paper to be something along the lines of an opinionated blog post. Initially I felt like being asked to comment on the syllabus was like being asked to comment on a bus schedule. There’s just not much to say. I had to work to come up with comments, and my comments all ended up being either typo call-outs or humorous asides.

• What did you think of Docker and operator overloading? (This question will vary, week to week.)

Docker is an amazing tool and an amazing feat of engineering. Earlier today I pulled a Docker image to try out a C++ interpreter wrapped in a Jupyter notebook. It worked flawlessly. I didn’t notice until later that the image hadn’t been modified for five years! For much of my career, the idea that you could get such a complex system running effortlessly, in a completely different hardware and software environment than it was developed for, after half a decade, was just unthinkable.

Operator overloading is a mixed bag. There are cases where operator overloads are justified. A good example is implementing + to work on vector and matrix types. These overloads allow us to write a program in a language closer to the problem domain than otherwise. So operator overloading can make code easier to understand.

But it’s easy and tempting to abuse operator overloading, especially in C++, and here’s why: we (humans) don’t like programming expressions like

print(print(print(print(print(cout, x), " + "), y), " = "), x + y);

For native readers of left-to-write languages, it puts the operations in the wrong order: the rightmost mention of print is the one that executes first. And we (humans) generally just don’t like deep recursion in our languages. (Try to understand “The girl the boy the dog bit hit cried.” I promise you it’s perfectly grammatical English.) So all those nested parentheses make us uncomfortable. (Pipe down, Lispers.)

We would much rather read this:

cout.print(x).print(" + ").print(y).print(" = ").print(x + y);

Ah, so easy to read! The order of execution matches the order of the code, and there’s not a nested parenthesis in sight.

But there’s a problem: in C++, you can’t add your own methods to a class unless you can change the definition of the class—and you can’t change the definition of ostream, because it’s part of the standard library. Maybe the standard library could put one print method on ostream for each built-in type, and one more taking anything that inherits some std::printable abstract class? Well, C++ needed an I/O API before multiple inheritance was straightened out (it’s still kind of a mess), so that wasn’t an option either.

So what is a young (35-year-old) language designer to do? Does he extend his language to allow custom operators? Does he allow adding methods to existing classes? Does he allow writing free functions infix (like we can do in Haskell, say x`mod`y)? Those would all be general-purpose features that would significantly improve the ergonomics of the language. But no! He picks two operators that look somewhat like arrows and have a precedence level not perfect but not hopelessly wrong, and overloads them to double as I/O operators. Fie, I say! Fie!

• What made you happy this week?

My daughter got perfect scores on her weekly Mandarin quiz.

• What’s your pick-of-the-week or tip-of-the-week?

Tom Preston-Werner’s “The Git Parable” is the best explanation of Git I have found. Reading it made me much, much better at using Git and at recovering when I messed up a repo or working tree.

See you next week!