Howdy, Stranger!

It looks like you're new here. If you want to get involved, click one of these buttons!

technologyThread

1234568»

Comments

  • It's not automated merges, its automated deployment after merging. You can still restrict merges into the master branch and make linting part of your deployment script
  • @toon_malk Yeah using multiple frameworks to keep track of the state of different parts of your app sounds like a nightmare, unless you could somehow efficiently sync the state between those frameworks and I highly doubt that's feasible.
  • We used Jenkins at my old job to essentially do the same thing
  • @toon @yama i've gotten really into normalization lately and am starting to look at frontends as fundamentally search clients all having their own structural preferences of denormalized data being indexed from a god machine.
  • edited August 12
    We use Jenkins at my current job, which I think is tied to a Git webhook for our master branch, and Peer Review is required to merge into master. I've only been working at my current job for a month, so I'm still learning their infrastructure.
  • god machine being some larger system that can take disparate datasources and map them into a relation schema that can actually be used by whoever is running the business.
  • @Ocaml sounds like GraphQL...?
  • edited August 12
    @yama hah yes resolvers are dope but i mean a system that generates gql schemas and manages the API from a higher level of abstraction.

    https://www.drupal.org/project/graphql
  • I almost implemented a graphql server in my side project, but I'm using firebase and their cloud functions are sooooo sloooowwww
  • i just use a docker node container with an apollo gql server on it and deploy it to whatever port 80 i have access to.
  • oh cool! what cloud are you running it on?
  • i was also looking at apollo as the server, but i wasn't sure how i'd integrated it into mobx on the client side
  • running it on ec2 instances.

    if mobx is using the same gql queries you defined then changing the services shouldn't matter.
  • @Ocaml right. that was the initial allure for me to move to it, and I might decide to move to it at some point anyway.
  • edited September 24
    Just upgraded to Mojave, I'm loving this dark theme.
  • this was great read, very informative and explained the concepts involved clearly and concisely. i do wish that the author explored the reasons that ConceptNet did end up being more effective, but perhaps that's for another article. I'm definitely interested in doing my best to read and understand the linked research papers.

    I appreciate the bit about pointing out that most people will take the path of least resistance. I think this is true to a significant extent, in my opinion people are not good or bad - but they are lazy. Therefore, it is crucial to create systems where making the good choice is also making the easy choice.

    I also appreciate that the author called out all the flawed assumptions in how this data is coded, and while it seems obvious (especially in hindsight) it is also true that most people working with AI are probably not taking such serious steps to determine their approach.

    "the upside is that you take the time to actually see what your data is doing. And that’s something that I think should happen more often in machine learning anyway." - I think this kinda drives the central thesis of the article home, we must understand what we make. The fact that this isn't even valued by most tech companies at the moment is extremely frightening.

    To believe in the perfection of what we (and I use "we" primarily in the humanistic sense) have created would be to believe in our own perfection. And if there's anything at all we can be certain of: it is that we are not perfect.
  • "An NLP system gets to learn from the best and the worst of what people say online, but the thing about the worst is that there’s more of it."
  • me n me pals are using an evolving neuro network to play zelda which is p cool
  • that's pretty rad
  • symfony is a great framework. super easy to pick up and comes with some really great data modeling tools. i feel like i can build anything with this framework. its all based on reusing services by injecting interfaces thru constructors and allowing configuration to swap out implentations.

    spent a while in the functional programming world in frontend development, especially state engines, and am starting to appreciate OO as server side cheat codes.
  • anyone have a fave hashing function

    my go to is sha1. its internal to php and its helping me resolve identity when processing potentially redundant data
  • also anyone fucking with ETL
  • edited October 12
    @Ocaml That injecting of interfaces through constructors sounds like dependency injection, correct?

    Sha1 isn't secure, but i guess collision detection could be a useful application of sha1.

    I wrote an ETL tool for my last job that took raw student test scores and imported them into our student information system. The first version was written in RxJS javascript, but it got to be too complex so I rewrote it in Python. My current job at a university uses ETL a lot, so I might be getting into that at some point in the future.
Sign In or Register to comment.