While I think it is technically possible, after a little bit of poking around, I decided that I didn’t want to bother trying. When I need an Airtable script that needs an OAuth connection, I have my script call a Make.com webhook and let Make handle the authentication.
How do you plan to create a new OAuth object when Airtable scripting doesn’t allow libraries?
Really? You know I really have other things to do that are way less fun than replying to you.
I’m not familiar with Replit. And it looks like I need an account to try it out. And that is just too much friction for me right now. The idea looks nifty. I’m also curious about trying out a coding co-pilot but not curious enough to shell out any money or sign-up for a limited time trial.
I agree with this. I think that low-code has amazing potential. It is also where I feel I belong. There also isn’t just one way of doing low-code. There are lots of ways to do low-code, just as there are lots of ways to do no-code or code.
I also think that there are lots of aspects of both writing low-code and learning to write code that I feel could use improvement. As surprising as it might sound, I miss some features of VBA. Using a coding co-pilot can help with learning to code and with writing code, but I haven’t used a coding co-pilot enough to give real feedback. Here’s my vague ideas based on my very limited perusal of coding co-pilot ads.
- I think that using a coding co-pilot could be very useful for identifying vocabulary issues. Don’t know what library to use or if one exists? Ask the co-pilot. Did you use
for ... inwhen you probably should have usedfor ... of. The co-pilot can flag that for review. - Just as grammar and spell check are useful for typing English, grammar and spell check are super useful for writing code. (Bill, I started using Grammerly on your recommendation, and I don’t hate it.) But I also think that memorizing spelling and grammar so that they become second-nature is important. Even if the co–pilot can fix spelling and grammar issues, writing code with fewer spelling and grammar issues in the first place will make for better/faster development.
- The co-pilot could also be useful for working with common patterns. I think this is most useful for experienced programmers who know how to quickly prompt the co-pilot to create code that fits a pattern.
- I suspect that a coding co-pilot could generate tests.
But my main issue is that I don’t think AI is ready to create complex logic for people who struggle to clearly articulate their needs. I think that novice coders really want this use case to work, and it just isn’t there yet. Maybe it will get there sooner than I expect.
What I really want is AI that can translate complex logic that I write in English into computer code. But I haven’t figured out how to do that yet. Maybe it is because I haven’t figured out how to write the correct prompts. Or maybe it is because the AI tools I’ve tried aren’t capable of doing it.
I also have vague ideas about using AI to generate a variety of training material, but I don’t have the bandwidth for experimenting with this anymore. I’m sure other people are working on this anyway. I think this has the potential to transform education.
Well, that’s a ramble.