This is an old project. I did the first version a few years ago and revisited it last year to make it use web workers. Under the hood, it’s a simple min-max algorithm, but on the highest levels, it’s tough to beat. I’ve learned a bit since I first wrote it, so I’d probably do it differently now, but it’s still a good demonstration.
…but maybe not learn software development.
Going back to around 2012, there was a movement to “teach everyone to code“, then there was a backlash. I’m going to come down right in the middle and say every one should learn to “code”, but not everyone should learn “software development”.
I once had to explain to someone (and a very smart someone I had worked with for years) exactly what my job entailed. They were surprised by my estimate I spent about 15% of my time as a software developer actively coding. Everything else supported that 15%, but it wasn’t writing code. Where did the other 85% go? Before I start coding on a project there is requirements gathering, research (mostly technologies and APIs) and design (figuring out how the code will be organized, not user interface). During coding, time is spent on unit testing, documentation and versioning. After coding, time is spent on integration testing, stakeholder acceptance and more documentation. And outside of coding, some time is spent as part of a team – meetings prioritizing what we’re going to work on, code reviews and supporting legacy code.
The point here is that software development is a lot more than just coding. And even coding is an enormous discipline. There are dozens of “common languages” worth learning, then there are data structures, algorithms, best practices, developer tools and the list goes on. You can never be done “learning to code”.
That alone may seem daunting, but the real reason most people should not learn software development isn’t that it’s hard, but that – outside of a career – it’s not that useful. Even “simple” programs can take hundreds of hours to write. And few people have the range of skills to write an entire program themselves. It would be like making a movie by yourself – you’d have to write, direct, star in, operate the cameras and microphones, edit and produce and handle distribution.
Ok, so if it’s that terrible, why should anyone – let alone everyone – learn to “code”. Well, coding changes the way you think, and much for the better. And not just how you think about computers, but about everything, at least everything involving information and decision making.
So spend some time in any one of the dozens of great resources learning to code. Plan on spending 100 hours over, say, a year. And you need to write some code, not just watch videos and take tests. Build something, check it in to github. When you’re done, you’ll find that even if you never write another line of code, you’ll use it all the time.
Ok, so maybe I’m late to the game on this one, but I’ve recently realized the value of including a clear command when I run tests. I started doing it bash a few months ago with
|
1 |
clear && go test |
Today I was working in windows and realized I could do the same thing with
|
1 |
cls & go test |
Of course, substitute “go test” for what ever language and operation you’re currently running. This significantly reduces the visual clutter and makes it much faster to iterate through debugging.
I’ll also include this here; My .bashrc when I’m working in Go includes these functions:
|
1 2 3 4 5 6 7 8 9 10 11 |
color()(set -o pipefail;"$@" 2>&1>&3|sed $'s,.*,\e[38;5;218m&\e[m,'>&2)3>&1 function g () { if [ -z $1 ] then clear && color go run main.go else clear && color go run $1 fi } alias gt="clear && color go test" |
The first, color(), is hard to read, but it changes the color of stderr. The rest let me use “g” as command to run main.go or “g foo.go” to run foo.go. And “gt” will run go test for the current directory.
I found a technique for decorating cakes a few years ago I really like. Start with a picture you want to use. Open it up in an image editor, I use Gimp, and convert it to two or three tones (I’ll add instructions later, but you can find a tutorial online). Reverse the image and print it off, making sure the size is approximately right.
Now the fun part. I tape the image on the back of a cookie tray, then cover the whole thing with plastic wrap making sure it’s pulled tight and secured in place (more tape). Then I use melted chocolate (white, dark and a mixture of the two) to trace the image. I put the chocolate in a sandwich bag and snip off a little bit of one corner to make a poor icing kit. You wouldn’t want to use a real one, it would be a nightmare to clean up.
The back side that you’re working on will end up looking terrible, but the side that is down against the cookie tray will come out, well, like the pictures above. The best part is, unlike fondant, people actually want to eat chocolate.
Good pizza sauce is magic. I started trying to make really good pizza at home a few years ago and this sauce is one of the keys. At it’s core, pizza sauce is tomatoes, garlic and oregano. Those three plus a little salt and you’ll have a better sauce than most store bought. The thing you really want to avoid is onions, that makes it taste like a pasta sauce. The rest of the seasoning I’ve played with over time and come to a balance I really like, but it will probably continue to change.
The cans should be 16oz. Try to find a sauce with no additives. Also, the petite diced tomatoes often have stuff added, get plain tomatoes. I like to use a wide, over sized pot. The large surface area lets the sauce boil down faster and the tall sides keep it from splashing.
Start by making a roux with the butter and flour, about 3 minutes on med-high heat, whisking the whole time. Add the tomato sauce and give it a quick stir. Then add everything else and let it simmer until it thickens, stirring occasionally.
I’m not sure how many times I’ve written a Mandelbrot explorer. But this is the best one yet! Drag to zoom in, the “Top” button in the upper left will zoom all the way out, and “Toggle” in the lower right will toggle the shading algorithm. You can also use browser navigation to go back.
The interesting thing about this project is that it was hand-coded in Asm.js. It was a good learning project to understand how Asm.js works, and using a validator was a big help. I plan on looking into Emscripten, but my first attempt to install it on an AWS Tiny instance failed due to low memory.