back to article From pair to p-AI-r programming: Kite floats paid-for spin of its GitHub-trained code autocomplete assistant

Kite will today launch a subscription-based coding assistant that tries to help programmers craft stuff quickly and efficiently. Founder and CEO Adam Smith told The Register his San Francisco upstart has spent more than six years developing a product that is, essentially, autocomplete for source code. Over that time, Kite has …

  1. overunder Silver badge

    Or...

    ... just take a typing class.

    You're supposed to write what is in your head, not what is suggested to you. If what is suggested to you is the bulk of what you'll write, you're rewriting too much. If anything, this "serivce" teaches you how to NOT write code as the less that is suggested, the better off you are.

    1. Glen 1 Silver badge

      Re: Or...

      Or....

      In ascending order of ridiculousness:

      Don't use intellisense, just take a typing class.

      Don't use Tab autocompletion, just take a typing class.

      Don't use regular expressions, just take a typing class.

      1. 2+2=5 Silver badge
        Facepalm

        Clippy for programmers

        It looks like you're trying to include a buffer over-run vulnerability in your code... would you like help with that?

  2. IGotOut Silver badge
    Terminator

    First....

    ....they came for the assembly lines...

    ...then they came for the delivery drivers....

    Now they are coming for the programmers.

    1. Ken Hagan Gold badge

      Re: First....

      I see nothing mentioned in this article that I'd distinguish with the title "programmer". As the first poster suggests, if your next line is that predictable, you shouldn't be writing it. You should be writing the bit that makes your product *different*.

      1. Glen 1 Silver badge

        Re: First....

        If that isn't programming, then taking it to its extreme, there will only be a handful of programmers left in the world. There will just be artists and mathematicians - and lets face it - most software is not exactly using 'new' mathematics.

        The 'hard' stuff isn't being done in bog standard line-of-business software shops. or web dev shops. or phone apps.

        1. MonkeyJuice

          Re: First....

          Just because you're not doing the 'hard' stuff, if I see 100 copy/pasted expressions with minor changes, I can assure you I am going to slaughter you during code review.

          Don't EVER write the same code over and over. If you can't abstract properly, you're chalking up technical debt that will choke even 'non hard' projects down the line.

          1. Glen 1 Silver badge

            Re: First....

            "I see 100 copy/pasted...."

            There is the well known rule of three.... Which this or similar software *would be able to pick up on*, and thus pass your code review on that measure.

            "Don't EVER write the same code over and over."

            Soo, by your logic we can only open 1 file, or only have 1 AJAX request, etc.

            I get your point about novel stuff requiring more brain cells, but yet another chart for management with tweaked parameters is not it. Yet another first person shooter is not novel. Yet another RTS is not novel. Yet another text editor is not novel - trying to deny that these tasks are not "programming" comes across as outright snobbery.

            How many careers have been spent entirely re-inventing the wheel re-implementing duplicate features in a competitors product?

            Sidenote: Do you *really* remember and understand how to do stuff, if you're only allowed to do it once?

  3. MonkeyJuice

    What a terrible idea.

    Intellisense is handy, it lets you trivially explore an API from the keyboard, helping you to remember that function name, or browse through and find functionality you didn't know existed.

    Emmet-like templates are handy for managing boilerplate (although much better to have sensible refactoring tools).

    I am perfectly capable of writing my own bugs. I have, in fact designed and implemented inductive programming algorithms, and the topic is an active interest of mine. I can assure you the state of the art does not even consider "throw it into a big pile of linear algebra and predict n-grams". That is stupid.

    They are doing it *completely* wrong.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2020