GenStage and Flow - José Valim (Lambda Days 2017)

Поділитися
Вставка
  • Опубліковано 7 бер 2017
  • www.lambdadays.org/lambdadays2...
    José will explore the rationale and design decisions behind GenStage and Flow, two abstractions that have been researched and now implemented in the Elixir programming with a focus on back-pressure, concurrency and data processing.
  • Наука та технологія

КОМЕНТАРІ • 14

  • @AlexanderBuch
    @AlexanderBuch 7 років тому +32

    Beautiful talk ;)
    Elixir really is crazy cool and José became my hero for all the work he has done!

  • @yogodoshi
    @yogodoshi 5 років тому +8

    So much enthusiasm, you can feel the love Jose has when he is speaking ❤️

  • @goofansu
    @goofansu 2 роки тому +2

    José is so good at explaining complex concepts, thank you!

  • @abulasarsayyad6839
    @abulasarsayyad6839 3 роки тому +2

    Wow.. just wow. How did I miss this thing till now.

  • @AndreasDonig
    @AndreasDonig 3 місяці тому

    Great talk. Thank you for sharing!

  • @seankemplay78
    @seankemplay78 4 роки тому +2

    Fantastic talk. Great communicator, great concept.

  • @haystackdmilith
    @haystackdmilith 5 років тому +2

    Glad I've been there :)

  • @sondiek
    @sondiek 7 років тому +2

    Great talk! Really cool design.

  • @elchemista
    @elchemista 7 років тому +3

    Great talk ! thank you

  • @arhyth
    @arhyth 6 років тому +1

    what’s the title of the paper Jose brings up near the end of the talk regarding concurrency on single machine, “Cost...” ?

    • @arhyth
      @arhyth 6 років тому +3

      found it linked from a plataformatec blog
      paper:
      www.frankmcsherry.org/assets/COST.pdf
      plataformatec article:
      blog.plataformatec.com.br/2017/03/small-data-with-elixir/

  • @dougb70
    @dougb70 3 роки тому

    the problem is that the difficult concepts are trivialized. Like how do you ensure that the "ares" are always going to the same node?

    • @jameshulse2
      @jameshulse2 2 роки тому

      Hashing

    • @50Kudos
      @50Kudos Рік тому

      The partition function is a mapper 42:30 and all producers know all consumers in advance. For example if your map function is just casting a value to boolean, the mapper is simply a simple case expression: [true -> state A, false -> state B]