Hey HN, my goal with Divjoy is to make it ridiculously easy to start a new React project. You just select all the things you want in your codebase (UI kit, React framework, auth provider, analytics, form handling, etc), pick a template, then export it as a complete codebase that you can keep building on. All the UI logic like routing and auth flows just works. It also has a built-in editor which I think is pretty nifty. You can view the entire component hierarchy right in the tool and drag in pre-built sections from the component library. I'm planning on building out a large library of components across all the major UI kits (Bootstrap, Material UI, etc) and adding support for Next.js soon.
This was by the far the most difficult project I've ever built. Would love to hear your thoughts and happy to answer questions.
This is awesome! I am a frontend noob and I've been learning a whole lot of things last 4 weeks and got a Vue SPA with Cognito auth, AWS amplify, Quasar and Vuex. I don't like the output I see in the browser and it pisses me off really. I am yet to cover UX, CSS/Stylus/SCSS and I see a bunch of bundle size warnings that I've no idea how to fix. For people like me, this project would reduce the bootstrap time significantly. I see you plan to add support for Vue etc but man, you're building something very valuable. Kudos.
I love the idea and I’ll use it !!! One question: apart Framework can we have a ‘Not needed’ option for all other features? What if I want authentication with no styling at all..... Thanks a lot and keep up !!,
Yeah, I'm definitely considering that. Basically just wanted to wait and see if anyone asked for it. Out of curiosity, is that because you'd like to use a different UI kit or do you generally write styles from scratch?
I think a lot of people prefer to write styles from scratch. At least I haven't seen these UI kits being used much around me. I tried using one a couple of times and while it's great for getting something on the screen quickly, it turned into hell when I wanted to customize the look. Also it emitted embarassingly large bundles.
I think most of the revenue would be from the small companies or agencies that need to make many sites fast on a regular basis. Once you have larger revenue then you wouldn't be using something like this — you would have your team.
Hey Chris, yeah Gatsby is awesome! I use it on https://usehooks.com. They're doing some really interesting stuff with themes, component shadowing, and of course their whole graphql system. Not totally sure how Divjoy would integrate with all that stuff (or whether it should try) but going to be exploring that soon.
Thanks for making such as awesome CSS Framework! Planning to keep growing the library of pre-built section built on Bulma, so hopefully it can become a great resource for the Bulma community (at least those that use React).
Classic 80/20 rule. I've used several of these in the past, and they always work great until you need that 1 thing that they don't support out of the box and then you have to spend half a day figuring out the JS hackery the builder uses so you can work around it to implement the one thing you need in it's subpar code editor that's hidden away in a submenu somewhere.
We recently started using another scaffold generator from http://scaffoldhub.io. It’s similar to this but geared more for apps than for marketing sites. It allows you to build out a complete data model and backend. It’s got a lot more features but they add complexity to the code that might not be needed for everyone.
Scaffoldhub looks cool. I can see how Divjoy looks like it's more geared towards marketing sites since that's pretty much what's in the component library right now. Going to be adding a lot of dynamic components over the next few weeks (think dashboards, social media feeds, activity dropdown menus, etc) and figuring out how to cleanly connect all this stuff up with backend options.
That's the proper thing to call it: scaffolding or boilerplate. "Codebase generator" sounds like an excuse to charge for the output it produces. Instead of that's just scaffolding I can imagine well, yeah, we used a codebase generator, that's just smart, isn't it? Should we do everything manually? Divjoy looks pretty cool but I think I'll avoid it out of principle.
Hey, is the main issue you have with Divjoy that it uses the term "codebase generator" or was there something about how it works that you didn't like? To be honest, I'm not a huge fan of "codebase generator" either. Open to changing it if it seems weird or meaningless to people.
That was just my reaction to it. I think it sounds catchy. I would just leave it and see how it works. I'm not the target audience because stock create-react-app and create-next-app (and rails new) are more boilerplate than I like for starting a project, though I'll look at custom boilerplate for inspiration.
Maybe a little of both haha? I mean it is ridiculous how many services you need to bring together for a basic web app. On the other hand, I like that the React ecosystem is composed of a lot of services that each do one thing really well. So not sure a big monolithic framework is the answer. So this was kind of my attempt at a solution.
This tool is an absolute joy to use. At my nine to five, I work on a library that often times feels clunky and heavy. To use a tool as simple, powerful, and beautiful as this one truly does bring me joy. Thanks for putting the time into this! It's the perfect intersection of drag-n-drop and from scratch.
If you've got some preferred channel where I can drop you a couple dollars in support of this project, let me know.
Wow, thank you! And I really appreciate the offer to support the project. The plan is to roll out some premium integrations and themes in the coming months and see how that works out as a revenue model. Not against setting up a Paypal or Patreon, but would love to see if this can be productized (yet still super generous with what you can do for free). Anything come to mind as something extra you'd pay for?
I think you're spot on with that revenue model. I would certainly be willing to pay a couple of bucks for a premium suite of templates and components, or integrations with less mainstream technologies.
One idea I'd really love is simple Git integration. If I want to download my code and work on it outside of the GUI, it seems cumbersome to return to the GUI for simpler tasks (e.g. let me double click this string instead of remembering which parent component has it). If I could sync to Git, you'd have all of my money!
As for a word of unsolicited advice: Don't be afraid to monetize! Plenty of people (and corporations) use and pay for tools like e.g. Figma which don't even produce code on this level. What you have right now is, in my opinion, more than generous for a free service.
To the author: Be careful going down the saas root. The path is filled with feature requests with promises of small monthly payments without lockin.
Great for a focused simple product but perhaps not so great for this kind of a tool. Ultimately your customers want something to create as much of the app/site as possible. As you approach this ideal complixity increases making your application more brittle. When the ecosystem changes your tool may not be able to respond to the new reality.
Your better off using this tool to create value for your own customers. Or selling the source code rather than chasing the 9 dollars a month customer.
Are you interested in something as simple as being able to export to Github or would you be looking for real git integration (with export and even import). I'm not sure on the feasibility of import but would love to hear more about what you have in mind there.
It's great to hear you find the code east to read and change. To clarify, would you want something like a rest API + node server that talks to an SQL database? The plan is to provide more self-hosted options for code export, going with 3rd party services like Firebase was just easier to add for this MVP.
Cool, thanks for clarifying. Definitely on the todo list, although still figuring out how soon I need to figure out the whole data layer (the big missing piece right now) VS focusing on expanding the component library across all the UI kits. Lots to do..
This is awesome, great work! One piece of feedback is that currently all of the components seem around designing static components. I'd love to see support for lists, dropdowns, etc, and ideally with the accompanying connect code to wire up a store with the renderer
Hey Martin, thanks for checking it out! Yeah, I'm eager to get a lot more dynamic components in there (lists, dashboards, social media feeds, dropdown menus, etc). Going to make that a focus over the next few weeks. These components will probably make it in before I have the data part figured out, but long-term want to find an elegant way to let the user pick from multiple data strategies (graphql, rest, etc) and just magically wire it up to the components.
This looks absolutely awesome. I had been toying with a similar idea in my head for a while, but I never even started building anything, and now it looks like I won't need to!
It had seemed to me that due to how web development has evolved in recent years, improving compositionality greatly by adopting functional programming concepts, designing something like this could be done relatively elegantly if you are just able to find the right abstractions. Seems to me that you have been able to do so, given what you support and are planning to support, though I can imagine it is an ongoing engineering challenge.
Kudos, thanks and good luck! I will definitely be trying this out, though I hope you decide to support typescript at some point.
Even the code sandbox integration is great, I dont even need to download the code. Good focus on what devs want...have you thought of integrating this with tools like VS Code ? More non web based product ?
Yup, my goal over the next month is to 5x the size of the component library, as well as start adding some more dynamic components like full dashboards, activity feeds, user settings pages, etc. Bootstrap is definitely coming. If you select it as an option you can enter your email. Just testing where the demand is. So far Boostrap and Material UI are about neck and neck. Feel free to shoot me an email at email@example.com if you want to get involved.
Thanks! So you can actually tweak SCSS variables like primary color in the tool. Just click the style button at the top when you have editor open. I'll have this for Bootstrap as well when it's added. Any kind of Vue implementation is probably a ways down the road, since the plan is to build some nice UI abstractions around React concepts (like Suspense). Harder to do that well if I'm building for multiple frameworks.
Thanks! Will likely try out premium themes and blocks. Maybe extra export options.
Accounts and multi-project support is definitely coming at some point. Right now you can technically import/export by clicking export project in the top dropdown menu (copies entire project json to your clipboard), save that, later import it using the import option. Not the most user friendly but it works.
Haven’t really thought about desktop to be honest! Will keep an eye on feedback and see how much people want that.
It was pretty exploratory at first so hard to say. Initially started out as me wondering if a drag and drop tree editor would be a decent way for beginners to learn React. Played around with a lot of different ideas and it kind of evolved from there. Maybe about 6 months full-time.
Hey, could you clarify what you mean by collapse/expand button for components? Currently, there is a +- button on each component that is a bit like collapse/expand. Technically it's controlling whether the component's inner structure is displayed inline in the tree or whether you just see the instance.
Hey David, I don't export anything that helps with server-side rendering the CRA app. In theory, I could do that if people want it, but my feeling is people that want SSR should probably just use Next.js once we support it. Coming soon!
This is helpful feedback. Right now it exports the normal CRA readme. Plan is to soon replace that with a dynamically generated doc that goes over your chosen stack as well as some details like the light wrappers we export around routing and auth. Is that what you had in mind? In terms of docs, anything else you’d like to see?