Thinking Outside The Box

Keynote for EuroPython 2013.

Transcript

1. ThinkinG
   - =
   outside
   The box
   

   View Slide

2. Hello, I'm Armin!
   I do Computers - with Python.
   Currently at Fireteam / Splash Damage.
   We do Internet for Pointy Shooty Games.
   c
   w
   

   View Slide

3. q
   t
   t
   j
   j
   d
   

   View Slide

4. the box is comfortable
   X
   

   View Slide

5. the comfort is dangerous
   l
   

   View Slide

6. the 9 dots puzzle
   0
   j
   5
   (and things of similar nature)
   

   View Slide

7. h
   - W
   <
   å
   ≥
   h
   ≥
   U
   using exactly 4
   straight lines,
   without retracing
   or removing one's
   pen from the paper
   connect the dots
   j
   

   View Slide

8. h
   - W
   <
   å
   ≥
   h
   ≥
   U
   using exactly 4
   straight lines,
   without retracing
   or removing one's
   pen from the paper
   connect the dots
   j
   

   View Slide

9. g
   i know it's cliché
   Ç
   

   View Slide

10. J
    the six thinking hats
    

    View Slide

11. J
    the six thinking hats
    This

    View Slide

12. the majority is always wrong:
    congratulations;
    you're now a cynical asshole
    7
    z
    “ ”
    

    View Slide

13. a change of environment
    1
    z
    m
    

    View Slide

14. c
    computer games
    

    View Slide

15. ˚
    4
    * this graph is not very scientific
    h
    ≥ Online

    View Slide

16. ˚
    4
    * this graph is not very scientific
    python!!!11
    h
    ≥ Online

    View Slide

17. community influences
    your thinking .
    a
    -
    ∫
    Î
    

    View Slide

18. C++ good;
    Scripting Languages Bad
    u
    

    View Slide

19. ... if it takes you 30 minutes to do a
    one line change then you obviously
    would not want to have runtime type
    checks ...
    E
    

    View Slide

20. it's too easy to dismiss
    something on fringe or
    outdated experience alone
    

    View Slide

21. • never underestimate how much your
    environment/community influences you
    1
    

    View Slide

22. • never underestimate how much your
    environment/community influences you
    1
    (unfortunately that also includes things like “the GIL is not a problem")
    

    View Slide

23. asking the right questions
    2
    j
    5
    

    View Slide

24. the wrong questions
    are easy to spot on others
    a
    -
    ˚
    4
    

    View Slide

25. how do I do something after return
    render_template(...) if I don't
    want to register teardown_request
    for all requests.
    J
    

    View Slide

26. • he was looking for celery / message queue
    ©
    

    View Slide

27. • first expectation was that the user wanted
    to change the HTTP response
    0
    

    View Slide

28. • teardown_request would not have worked
    anyways
    O
    

    View Slide

29. I don't want my user to wait while I do
    some processing on his data
    v
    “
    ”
    better:
    

    View Slide

30. t
    j
    the box was Flask
    

    View Slide

31. how to ask the right
    questions2
    -
    5
    B
    

    View Slide

32. • assume you already started out wrong
    • describe the •actual• problem
    =
    

    View Slide

33. How do I use Websockets
    with Flask?
    z
    “
    ”
    

    View Slide

34. How do I notify my user about
    changes with low latency.
    6
    better:
    “
    ”
    

    View Slide

35. 5
    question
    leaves room
    for the answer
    

    View Slide

36. (a) Server-sent events via WSGI
    (b) application <-> redis <-> persistent
    connection server
    Ô
    bit.ly/pypush
    

    View Slide

37. f
    Jump on IRC
    help out other people
    < >
    

    View Slide

38. questioning the right things
    3
    z
    m
    

    View Slide

39. the worst parts in my libraries are the ones
    where I took the design from elsewhere
    O
    

    View Slide

40. it's not because
    I know better ...
    ∆
    Ò
    

    View Slide

41. • most things have some design behind
    • as people copy it, the original design
    gets obscured and forgotten
    • the original design might no longer apply
    1
    

    View Slide

42. • starting something new?
    • question everything!
    ˝
    

    View Slide

43. not with the intention of proving existing
    design wrong;
    with the intention of understanding it.
    9
    

    View Slide

44. 9
    paradigm shifts
    4
    F
    

    View Slide

45. Many times we don't even realize that things
    were an example of thinking outside the box.
    (a) "echo" -> Request/Response objects
    (b) Interactive Interpreters
    9
    

    View Slide

46. That's also what makes it hard to find
    examples now ...
    9
    

    View Slide

47. every idea is a rehash
    don't get too excited when you feel all "obviously ..."
    sometimes all that's necessary is transposing a concept
    from one industry to the other.
    9
    

    View Slide

48. 9
    interesting examples
    5
    F
    

    View Slide

49. • Mill Processor:
    • Basic Block: One entry, one exit.
    • Break instruction bundles in half
    • Two decoders, one moves left from EBB
    entry point, one moves to the right
    =
    

    View Slide

50. • Mill Processor:
    • Basic Block: One entry, one exit.
    • Break instruction bundles in half
    • Two decoders, one moves left from EBB
    entry point, one moves to the right
    =
    two independent units, two separate caches
    

    View Slide

51. =a
    High-level Queues for Request Handling
    Stateful Frontend Servers
    Stateless Workers
    

    View Slide

52. • Sending around signed data
    • Cookies
    • Access/Refresh Tokens
    • Activation Links
    • Offers
    ˆ
    

    View Slide

53. • The Rust Programming Language
    • Memory ownership tracking
    1
    (and otherwise just steal from C++, C, Python, Ruby, Haskell and Scheme)
    

    View Slide

54. • Spotify's Native/Web Bridge
    • spawns HTTP server on localhost:XXXXX
    • provides OAuth bridge
    • JavaScript authenticates with local
    server, sends commands and retrieves
    updates.
    c
    

    View Slide

55. That's it;
    Now ask questions
    ?
    @mitsuhiko
    lucumr.pocoo.org/talks
    gittip.com/mitsuhiko
    y y
    y
    d
    

    View Slide

56. calibration slide
    --
    2 2 2
    2
    2 2 c
    

    View Slide