Good morning!

Today's a good day to ship something new.

As you may know, I've been involved in RSS as a standard way to distribute news since the beginning. I wrote the RSS 0.91 and RSS 2.0 specs, and shipped the first software that was compatible with both formats. My early products, My.UserLand and Radio UserLand got the market started, along with a deal we did with the NY Times in 2002 to get their news flowing on the RSS net.

The RSS market is slowly rebuilding after the dominance of Google Reader was followed by its closing a couple of years ago. It shows how resilient open formats and protocols are. If Google Reader had been built on proprietary formats, that would have been the end. But RSS isn't something anyone can own, an idea that confused many big companies. This is something they just don't understand, at a basic level.

As you may know, I switched over to development in JavaScript a few years ago, and one of the first things I did when I got to Node.js was build a RSS aggregator, River4. I built it for Heroku, which I thought was a revolutionary platform, because of its pricing and architecture. It was designed for lots of little apps that are easily deployed. I love that idea, it's one of the foundations that my own environment, Frontier, was based on. River4 was a good fit for Heroku, but that made it awkward to deploy elsewhere. Over time we took care of that, but that made it more complicated to configure.

I also learned about various ways to configure Node apps as I learned more about Node, and that is also reflected in the design of River4, and not in a good way. There are some things that can only be configured in environment variables, and others that can only be configured through a config.json file. It became a bit of a hairball.

So I decide it was time to do a restart of my JavaScript RSS aggregator, and it's now ready for Node users -- it's called River5.

It's open source, MIT License.

There's a mail list.

Example sites.

A Hello World app.

This ia a foundation for developers to build on, but it's also possible for an adventurous user to set up their own rivers. Ken Smith, an English professor, is running River5 on his personal computer. It works. ;-)

So what's new in this river?

It's faster since it builds the rivers incrementally as it scans the feeds. There's no intermediate database.

In addition to the standard list format, I've also defined a JSON format for lists and a plain text format, to make it easier for people to get going.

Most important to me, it's easier for me to work on, so there's a better chance of new features in the future.