Skeleton vs Hydrogen Demo for Template?

[osseonews]

I'm wondering what the best way to build a store is. I was considering using Skeleton and just installing Tailwind on our own. I then noticed Hydrogen Demo, but saw that it was not moved to separate repo. My question is that the Skeleton seems to get more updates and more frequently, so it seems like developing off of Skeleton is better. Do you plan to keep the Skeleton and Hydrogen Demo in synch so the Hydrogen Demo will always get the latest Skeleton update?

[jamalsoueidan]

It's very challenging to keep everything synchronized. I have been developing a hydrogen store, starting with a basic skeleton. Later, I decided to use Mantine UI. As time progresses, your code will begin to look different, and you can't just upgrade to latest version.

Every three months they release new bugs and features, then I copy my code into a new skeleton just to upgrade to the latest version

My code now use the old account api, the latest skeleton use the new account api, I can't just upgrade, I must get a fresh skeleton and apply my code on top of the new skeleton.

I'm waiting now for april release so I can do that, it take me more or less some hours to do that.

[osseonews]

Thanks, yeah, that's been my exact experience with "headless" commerce now for a few years and why I'm starting to tire of it a bit, but giving it one more shot... You begin with some template and then it diverges so much at a later date, the maintenance becomes a nightmare and so much extra work. For a few projects I just gave up updating anything. I was hoping they would just standardize on some basic set up/design, like the Hydrogen Demo Store and then I'd try to develop around that in some fashion that made updates easier.

[blittle]

The skeleton template is a more appropriate starting point for a new store. It less opinionated than the demo store. It's what you get when you run npm create @shopify/hydrogen@latest. The demo store is much more opinionated with a specific design. I recommend using the skeleton template as your starting point, and the demo store as a reference for more advanced features.

[osseonews]

So I spent the last few days coding with the Dawn theme in Liquid. What a pleasure. It's quite easy to customize anything with a bit of javascript and CSS. Updating the theme? A breeze.

Hydrogen/Remix is precisely the opposite. I'd say it's nearly impossible to maintain, because by the time you finish building, they have already changed so much in the core, you have to build many things again. You end up wasting tons of time just fixing things that should work out of the box.

From my perspective the only way Hydrogen will ever work for Shopify is if they actually have a starter theme that has everything , is constantly maintained, and looks good. Sort of like a Dawn theme for Remix. Then they should have a "plugin"/"swizzle" type system to override core templates so you can easily customize things without fear of missing out on new updates or having to refactor every few months. They should take a look at how Docusaurus does with swizzling and plugins. It's a great system. Yeah, you are somewhat constrained in terms of design under this type of framework (most Docusaurus sites look the same), but that is really irrelevant nowadays, because consumers expect e-commerce websites to look a certain way now and there is no benefit in looking that much different. So if they just had a decent e-commerce theme with Tailwind, it would be fine for 90% of cases, and when you need a customization you can "swizzle" or develop a plugin. Right now, by the way, the Demo Store Tailwind theme is missing alot and I could never just start with that.

[jamalsoueidan]

You need to understand that the Hydrogen framework is for people who want a lot of customization. It means you should develop it yourself. The theme is suitable for stores that simply need to sell products and customize the design. If you need to add extra functionality, you can do so with Shopify apps; just install them, and they will modify your theme. I did that with my last store for 6 years, if I had hydrogen it would had made my life easier.

Hydrogen is perfect if you want a custom store, like the one I built. I'm using Shopify metaobjects to implement a headless CMS on top of Hydrogen, and my own backend to add more customization.

They had a Hydrogen store with Tailwind, but they removed it in the last release. Perhaps they will bring it back if many people request it.

[osseonews]

Sure, I understand that. I've been building custom e-commerce stores on React for 4 years already before Hydrogen was even a thing. Built on Gatsby+Magento, and Nextjs/Magento etc. Hydrogen itself has already gone thru a major revamp, from when they bought Remix. Anyway, my point is that you will not get very much from customizing on Hydrogen now vs what you can get on Liquid now, and the maintenance will be hell on Hydrogen. If you can program in Remix, you can make an app on Liquid and use your own backend.

Now I'm not saying Liquid is the future. I certainly don't think so, and would love to do everything in React with Remix or Nextjs. But, it's silly to spend time building core features, like Customer Auth, when everything should just work out of the box. And who really wants to spend time styling various components to build what in the end, will basically look the same as every other ecommerce store. Most successful e-commerce sites all have the same basic design b/c that's the way everything works at some point, there is just a standard flow that customers expect. Finally, it's nuts to have to refactor stuff so often, because the underlying code is changing so often. I'm certain you didn't make a custom store to have it stuck in time. Presumably you will want to add features and keep up with the latest upgrades from Shopify. This won't be easy and the payoff won't possibly be worth the effort, in any monetary sense.

So what I'm saying is that if Hydrogen is to have a future, they need to develop a good-lucking starter theme, a Dawn for Hydrogen and the have a Docusaurus-like "swizzle" and "plugin" architecture so upgrades are easy and new features are easily deployed to existing set ups. Then, of course the customization effort will be worth it, because you can feel confident to customize where needed, knowing that you still easily take advantage of new features by a simple pull.