I expect the majority of my projects to fail, and they do. As a rule, off-time projects are entirely open-source, and ship early and often. There are certainly those who disagree, with this sentiment, but it’s a sort of personal preference at the end of the day.
Big was announced 2011/10/28, 348 days ago. Since release, Brian Campbell contributed better history management, mdznr created a fork that uses Markdown, golden ratios and some other stuff. Jed Schmidt made weenote, which looks great and is arguably even more code-golfy.
simple-statistics was introduced 2012/06/26, 105 days ago. jthomm and Matt Sacks contributed bugfixes and features, including support for the student’s t-test. It now has 69 stars on GitHub but no dependents on npm.
Happen was announced [2011/10/18](2011/10/18](http://macwright.org/2011/10/18/adventures-in-testing.html), 358 days ago, and has gained keyboard event support thanks to Matt Tortolani. I’ve also added tests and bugfixes for new corner cases. 20 people follow the project and it’s used by Modest Maps for testing.
So far they haven’t had any contributors. Which is fine with me; I find these to be really important projects because of nothing more than the scarcity of open-source Objective-C. Just having the source for minute around when I was writing heard cut development time in half, and I honestly think they might do the same for another developer.
Having a great
README.md is key to making concepts concrete: it should
explain the project in full, with no website or blog post to necessarily back it up.
It’s also vital to have
LICENSE from day one, and I’ve found no reason
not to go BSD for everything.
The impact of testing on usage isn’t super-clear. The Travis-CI badge looks great and provides some sort of stamp of approval, but its false negatives are irritating and untested projects like Big seem to succeed in their own right.
GitHub is Rome for software, but not for communications; you need to run communications from Twitter and ideally a website in order to make things accessible and clear. This is a part of software development has been one of the most unusual; you can’t really be silent. People do not spontaneously find out about things: big MapBox projects have been open on GitHub for weeks with no attention until they become big news with a blog post.
Because writing software takes time and effort. As a rule, all software I write outside of work is free & open source, but it isn’t free to generate - it’s time that could be spent doing other things. And on the other side, it’s import that it has value, which is basically the aggregate of its usage and all of the things that means.
For now, there are about 30 ideas in my ‘secret plans’ Notational Velocity note waiting to happen.