Twidi, I have one more question. Are you caching all the github api data on your end or do you hit the api directly and assume you are good for the limit?

Everything is stored by my side. All API requests are asynchronous, when needed only (a fetch of a repository cannot be done if it was done in the last 30 minutes, and so on)

(and i forgot to say that repos.io add "main" tags automatically, for repositories and users)

twidi , no problem, no one knows about http://narly.us because I have been building it in a vacuum =/

Perhaps we can help one another? Through narly is in ruby and repos.io is in python. In any case would be great to collaborate. My code is also fully available : https://github.com/plusjade/narly

Do you have any experience with redis or backbone.js ? I think backbone.js is supercool (play with the narly tagging ui if you can) but it does have a non-trivial learning curve and a lot of closure/reference gotchas that I struggled with. Anything I can do, just let me know!

Yes why not, it sound like a great idea.

I love redis, use it a lot on other projects, and it's used on repos.io but not for main storage (postgres and solr).

I don't know ruby and backbone (and really i can't read ruby !)