Misframe

Fun to obligation.

Published Jul 27, 2014

A few months ago I was sitting in the back seat of a minivan on the way to New Jersey to visit family. I had music playing through my headphones and I was staring at a bunch of sFlow specification documents that I had downloaded onto my laptop. sFlow is a packet and counter sampling specification, and I wanted to write a decoder in Go.

It’s pretty neat to have the entire history on GitHub. I had my first interesting commit pushed at 9 PM on a Sunday. You can browse through the commits to see how I went from a prototype that barely did anything, to a much higher quality Go package. I worked on it in my free time (because I have so much :P).

I sometimes tell people about sFlow and my decoder, but I find that no one knows what it is. I guess it makes sense since sFlow usage is usually limited to like… network engineers and the like. Even the network engineers I know aren’t really programmers. So yeah, no one is really using my package or talking about it other than me. It’s just a personal project, and I like to work on it whenever I have time. It’s largely incomplete simply because I don’t need the rest of the features, but I still use it and make fun little tools for demonstration.

The other day I got a notification1 about a pull request: https://github.com/Preetam/sflow-go/pull/10. It was a bug fix, and the person seems to be from Turkey and works for an ISP. I was actually shocked, and for two reasons. First, I didn’t know that people would actually use this personal project of mine. Cool! And I didn’t know it was buggy! I didn’t notice anything wrong, but I haven’t tested it much (it is a personal project, after all).

I saw the email in the morning so I had to wait until after work to take a look. I think in general, if I see something wrong with one of my projects I just make myself a reminder and fix it later. It was different this time because I knew someone was expecting a fix :P. I went to sleep at like 3 AM after finding a fix.

I spent a lot of my free time working pro bono to fix this. Turns out it’s still not fixed! Again, I need to spend time trying to fix it for nothing in return. Feels a lot like an obligation at this point.

Something I started for fun to pass the time is now an obligation eating up my time. Interesting. I also think it says something about refinement. Figuring something out and creating something new always seems to be fun, but refinement seems tedious. But hey, I think I’m becoming a huge fan of nitpicking.

By the way, I have high expectations for this package. The only other implementations of the sFlow protocol that I’ve seen are written in C and Perl. I can’t imagine anyone writing a high-performance, “cloud”-ready flow analyzer using C or Perl. If you see one, let me know! Go, on the other hand, seems much more suitable. I want this package to be the most elegant (and performant) sFlow decoder (and encoder, hopefully :P) implementation out there.


  1. By the way, GitHub’s filtering is either broken or too complicated. I have no idea why I see organization events on my personal newsfeed.