Get the latest tech news

The success and failure of Ninja (2020)

Success and Failure of Ninja Around nine years ago I published Ninja, a build system that is mostly comparable to Make. At the time I was a bit embarrassed to share my side project but since then it has become widely popular.

Ninja's insight (discovered in retrospect) is that all of these tools, no matter the high-level features, ultimately eventually must construct some sort of graph of the actions: the files they intend to keep up to date and which commands to execute. On the other side, it also meant Ninja was useful in very flexible ways, because the generator could be as high level as the user wanted ("tests are found by globbing the entire source tree for files with 'test' in their name"). Another way of saying this is that today I am motivated by just trying to impress or live up to the ~ten hackers that I admire, people like apenwarr or agl or bradfitz or graydon, and while it's occasionally cool to meet someone and have the reputation of my software precede me, I think a lot of "succeeding" was mostly just kind of a burden.

Get the Android app