Future plan for nbviewer.

Finish nbconvert, probably have it merge into ipython or self installable.

• Think of the url scheme, and the services we want to expose.
• Think about caching/ and the frameworks we want to use (keep flask? look at Flask Blueprint.)
• Open a google analytics account ?
• Do we start storing data about users ?

• 0 - Fix the annoying bugs (404/500 cached)

• 1 - move to mongodb (free postgres is too limited)

• 2 - Use the new NBconvert class

• 3 - Write a new jinja template for nbconvert.

  • that does only include notebook body
  • include the notebook css in the header of nbviewer itself

• 4 Include boostrap menubar with different actions

  • share on twitter, fork, share on G+

• Licence in footer (nb extension + metadata)
• Authors in footer (nb extension + metadata)
• Use twitter "cards" and Google+ "snippet"
• Integrate small screencapture tool using my boiling-chamber heroku app
• List all user gists that are ipynb files
• List the more popular/viewed ipynb files

• subdomain vs url for different services :
  • more robust to faillure.
  • independant update
  • load balance traffic
  • less expensive
  • Could we share some cache and data ?
• url scheme :
  • / main page
  • /[prefix]/, some [prefix] should be reserved at our usage.
    • static
    • faq
    • contact
    • url
    • urls
    • help
    • github?
    • gist?
    • others?
  • /[numbers], keep for backward compatibility with before
  • /u/[username]/ list of user gist
  • /g/[username]/ list of user repos (browsable)
  • /u/[username]/number corresponding gist if tree, browsable if blob
• some place to download bookmarklet/notebook extensions