I'm currently using Python with Django and PostgreSQL on a VPS, but I'm on the lookout for a more modern web framework or language that's optimized for AI-assisted coding, especially with language models in mind. I'm hoping to stick with PostgreSQL without diving into custom setups or Docker. Ideally, I want an open-source option driven by community input rather than a single developer who might change things without notice.
I'm leaning towards TypeScript-like languages, which provide stricter syntax than Python and eliminate indentation issues that complicate code injection and pasting. The aim is to reduce boilerplate code as much as possible while maintaining context. I'll continue using Python and Django, so please avoid recommending those.
I'm considering a few options:
- Wasp, which seems promising but is still in beta.
- Encore.ts, which is heavily AI-focused, but I'm unsure about its deployment approach as it appears to be a single company project.
- AdonisJS, which feels a bit outdated yet allows flexibility, though it's less AI-oriented.
- Django remains a solid choice, though it has considerable boilerplate.
Has anyone here worked with these options and can share insights or recommendations on this topic? Let's keep any negative views on AI out of this discussion!
4 Answers
For what it’s worth, I believe Django still has its merits especially with its built-in tools. But yes, the heavy boilerplate can be overwhelming at times. And because of its older style, it can sometimes be cumbersome when trying to integrate AI tools effectively.
Since I have a background with Django, I think it’s worth noting a few strong points for those considering it:
- The admin interface is incredibly helpful as it’s auto-generated and can save you valuable coding time.
- Modularity is a plus, as 'startapp' allows you to compartmentalize functionality which helps maintain context and reduces project bloat.
- Moreover, Python is one of the most recognized languages and has a vast amount of code and practices available, which can be useful for AI models.
However, the trade-offs include the enforced indentation that complicates editing workflows, as well as a myriad of ways to accomplish the same task leading to potential confusion.
Among the frameworks you mentioned, AdonisJS could be a solid long-term choice. Wasp looks efficient but is still quite young, while Encore.ts gives me a little anxiety about being too dependent on a single vendor. For AI-assisted coding, I'd prioritize conventions that are straightforward, strong TypeScript types, and a lot of documentation. AI models tend to perform better with more universally accepted patterns, rather than the ones marketed as 'AI-native'.
I've been using Next.js with PostgreSQL through Drizzle ORM for about a year, and it seems to meet your needs. TypeScript helps avoid indentation hassles, and with a vast ecosystem, you’ll find many resources available. In terms of Wasp, I found that the DSL layer can complicate interactions with models since it requires understanding two frameworks. AdonisJS is good, but a smaller community often translates to fewer examples during development. I’d recommend leaning towards Next.js or Remix, as they have enough momentum to align with what AI models are already trained on.
Exactly! Using Next.js paired with ORM tools and libraries for front-end might be the way to go for single-use apps or problem-solving tools. While I didn't suggest it originally since it can vary greatly, that approach aligns perfectly with how an AI would generate a full-fledged app component. For consistency in deployment and updates, I prefer a singular solution too. I've dabbled with Flask and Bottle, but managing multiple small repositories just became messy for me.
You make some great points! I've had plenty of stability with Django over the years. AdonisJS seems to mimic Django but in TypeScript, which is tempting. Still, I'm not sure it's worth switching just for that. I'll definitely give it a try for prototyping.