QCon 2007 Notes: Zed Shaw on Mongrel

Zed Shaw, Mongrel, 2500 Lines, and Economics

Partly about Mongrel, partly about what makes a wildly successful open source project.

  • The key to Mongrel having a hundredfold better uptake than anyone else: documentation right on the home page
  • Do frequent releases, but only after a few people agree that it’s ready. Release too early before that, people will leave in droves.
  • Give answers, not explanations
    • If you have to keep explaining something, maybe you should fix it so it works the way people expect. This is pure frustration and economics.
  • The really successful things are really well documented. And that’s “how to use it”, not “how it works”
  • Ultimately, it’s all about removing suffering. Make your product easier to use than the competition.
  • Stuff to do to enable you to get stuff done in open source with just a little bit of pain:
    • Grow a sense of humour and a thicker skin
    • Be symbiotic rather than parasitic
    • Give stuff back (for instance, well-written documentation is incredibly important)
  • Mongrel: Problems to solve (or things that the Mongrel-beater should fix)
    • Clustering
    • Sessions (right now, it uses cgilib, which is horrible and buggy)
    • Caching
    • Logging
    • Recovery
    • Management
    • Actually, even just Mongrel with better gc analysis would be a big win
  • Mongrel: Things not to change
    • One minute start (just three commands: immediate gain)
    • No options required (works out of the box)
    • Strict operation
      • this actually and inadvertently blocks some security attack: most clients have to try to behave, because they don’t know what servers they’re talking to: the clients that don’t are usually the ones trying to break things
    • Good Documentation
    • Processes not Containers
    • Simple API

It's only fair to share...
Share on FacebookGoogle+Tweet about this on TwitterShare on LinkedIn

Leave a Reply