this post was submitted on 15 Aug 2023
91 points (89.6% liked)

Technology

59578 readers
6091 users here now

This is a most excellent place for technology news and articles.


Our Rules


  1. Follow the lemmy.world rules.
  2. Only tech related content.
  3. Be excellent to each another!
  4. Mod approved content bots can post up to 10 articles per day.
  5. Threads asking for personal tech support may be deleted.
  6. Politics threads may be removed.
  7. No memes allowed as posts, OK to post as comments.
  8. Only approved bots from the list below, to ask if your bot can be added please contact us.
  9. Check for duplicates before posting, duplicates may be removed

Approved Bots


founded 1 year ago
MODERATORS
 

ChatGPT Isn't as Good at Coding as We Thought::undefined

you are viewing a single comment's thread
view the rest of the comments
[–] abhibeckert 6 points 1 year ago* (last edited 1 year ago) (1 children)

So this isn't a real example, it's just something I slapped together now as an example of how I generally use ChatGPT 4.0. In a more realistic scenario, I'd be asking more detailed questions, pasting in my existing code, and asking the AI to write smaller sections (maybe 10 lines of code at a time).

A single chat might run for five or six hours, continuously discussing an individual task I'm working on, and I won't just be asking it to write a bunch code. For example in this case I might spend 10 minutes hashing out the exact contents of the 'Activity' record (or, maybe I already did that a month ago, and I might just paste in an example). I'd also be testing the code as I go and writing some of it myself, occasionally asking questions like 'how do I format a date as 4 Feb 2016' in JavaScript?' or 'this line of code fails with X. Why?'

In those five or six hours, I estimate I'm able to get two or three days of work done (at my level of productivity before I started paying $20/month for ChatGPT+).

I'm also only pasting the code - ChatGPT also explains the code it outputs, and often those explanations are more useful than the actual code.

In JavaScript write a "reserveSeat('A42', 'Adult')" function that sends an ActivityStreams 2.0 activity to (this.apiUrl)/reserve and inform the user if it failed to be reserved

class SeatReservationService {
  constructor(apiUrl) {
    this.apiUrl = apiUrl;
  }

  async reserveSeat(seatNumber, seatType) {
    // Construct an ActivityStreams 2.0 compliant JSON object
    const activity = {
      '@context': 'https://www.w3.org/ns/activitystreams',
      'type': 'Reserve',
      'actor': {
        'type': 'Person',
        'name': 'User'
      },
      'object': {
        'type': 'Seat',
        'id': seatNumber,
        'category': seatType
      }
    };

    try {
      // Send the activity object to the specified API endpoint using an HTTP POST request
      const response = await fetch(`${this.apiUrl}/reserve`, {
        method: 'POST',
        headers: {
          'Content-Type': 'application/json'
        },
        body: JSON.stringify(activity)
      });

      // Handle the server’s response
      if (response.ok) {
        // If the request was successful, log a success message
        console.log('Seat reserved successfully.');
      } else {
        // If the server responds with a status code other than 2xx, throw an error
        throw new Error('Failed to reserve seat. Server responded with ' + response.status);
      }
    } catch (error) {
      // Inform the user if the reservation failed
      console.error('Failed to reserve seat:', error);
      alert('Failed to reserve the seat. Please try again.');
    }
  }
}

// Usage
const service = new SeatReservationService('https://example.com/api');
service.reserveSeat('A42', 'Adult');

Write a PHP request handler that does $this->reserveSeat('A42'), and checks if it failed with a SeatNotAvailable exception

class SeatNotAvailable extends Exception {
    public function __construct($message = "Seat not available", $code = 0, Exception $previous = null) {
        parent::__construct($message, $code, $previous);
    }
}

class SeatReservationService {

    public function reserveSeat($seatNumber) {
        // Mockup condition that the seat 'A42' is not available
        if ($seatNumber === 'A42') {
            throw new SeatNotAvailable("Seat $seatNumber is not available.");
        }

        // Normally here you'd interact with a database or another kind of storage
        // to actually reserve the seat.
        // ... (reserve seat logic) ...
        
        return true;
    }
}

class SeatReservationController {

    private $service;

    public function __construct() {
        $this->service = new SeatReservationService();
    }

    public function handleRequest() {
        try {
            $this->service->reserveSeat('A42');
            echo "Seat reserved successfully.";
        } catch (SeatNotAvailable $e) {
            echo $e->getMessage();
        }
    }
}

Write a PHPUnit test for SeatReservationController, and a Jest one for SeatReservationService

createMock(SeatReservationService::class);
        $serviceMock->method('reserveSeat')
                    ->willReturn(true);

        $controller = new SeatReservationController($serviceMock);
        $this->expectOutputString("Seat reserved successfully.");
        $controller->handleRequest();
    }

    public function testSeatNotAvailableException() {
        $serviceMock = $this->createMock(SeatReservationService::class);
        $serviceMock->method('reserveSeat')
                    ->will($this->throwException(new SeatNotAvailable()));

        $controller = new SeatReservationController($serviceMock);
        $this->expectOutputString("Seat not available");
        $controller->handleRequest();
    }
}
const SeatReservationService = require('./SeatReservationService'); // Adjust the path as necessary

test('reserves a seat successfully', async () => {
    const service = new SeatReservationService('https://example.com/api');
    global.fetch = jest.fn(() => 
        Promise.resolve({
            ok: true
        })
    );

    await service.reserveSeat('A41', 'Adult');
    expect(fetch).toHaveBeenCalledWith(
        'https://example.com/api/reserve', 
        expect.objectContaining({
            method: 'POST',
            headers: {
                'Content-Type': 'application/json'
            },
        })
    );
});

test('fails to reserve a seat', async () => {
    const service = new SeatReservationService('https://example.com/api');
    global.fetch = jest.fn(() => 
        Promise.resolve({
            ok: false,
            status: 400
        })
    );

    global.console = { error: jest.fn(), log: jest.fn() };

    await service.reserveSeat('A42', 'Adult');
    expect(console.error).toHaveBeenCalledWith(
        'Failed to reserve seat:', 
        expect.any(Error)
    );
});
[–] TropicalDingdong 6 points 1 year ago

In those five or six hours, I estimate I’m able to get two or three days of work done (at my level of productivity before I started paying $20/month for ChatGPT+).

This is very comparable to the ROI I would say that I've been seeing for my programming work. I feel like a super hero, or a 22 year old on adderall. I know everything I need to do for any project, but between client meetings, executive meetings, business development, meetings with the product team, mentoring, the actual amount of focused time I get is so little. I can offload a huge amount of the "I know how to do this and I'll know if you do it wrong, but please do this for me" to the machine. This past May I took on a task that would have taken a comparable person, probably 6 months, and I knocked it out in 2.5 weeks. If you already know what you are doing, ChatGPT is steroids.