payload VS Grav

Despite being a relatively newer player, Payload's GitHub repository has accumulated 18.8k stars and 1.1K forks as of April 2024, reflecting its growing community. The project has also secured $5.6 million in funding, positioning it for continued growth and innovation.

My most recent project launched in January. NextJS 14 client integrated with PayloadCMS (http://payloadcms.com) for the back-end. I love both technologies in theory, but they're both going through a renaissance period and "bleeding edge" doesn't even begin to describe it.

If I'm just building a client app, create-react-app is still my go to.

Before now, I'd been building on WordPress for 10+ years for anything client-administered. Planning on using Payload from here on out.

Payload CMS: The Customization Insurgent

Payload is free, you can self host it without paying a one time fee or a SaaS fee for its use, it even says so at the bottom of the homepage

James, the co-founder of Payload, a headless CMS with MongoDB support, shared his insights on the drawbacks and limitations of using a headless CMS in the context of web development. He challenged the promises often made about headless CMS, such as separation of concerns and ease of content migration, revealing that these claims often don't align with the reality faced by developers and clients. James is considering integrating Payload directly with Next.js to overcome these limitations and offer a better developer experience, including out-of-the-box features and simpler deployments. Should Payload move to Next.js?

Hey HN, Dan here from Payload (YC S22), an open-source headless CMS that closes the gap between CMS and traditional app frameworks. We’re excited to announce Payload 2.0!

https://github.com/payloadcms/payload

If you’ve not heard of Payload you’re probably wondering why the world needs another CMS. Payload connects to your database and runs without the vendor lock-in and black box of SaaS based CMS solutions, and it’s far more extensible than off-the-shelf SaaS options. Enterprises in specific have been finding value in this control, and they’re using Payload to power content infrastructure that simply isn’t possible through integrating with SaaS webhooks alone.

Today’s announcement is all about features that strike at two neglected areas in the world of CMS. The first is application framework level control over your database that you’d expect with tools like Ruby on Rails or Laravel and the second area is making content editors effective by seeing their edits in realtime.

Here are the highlights on what we’ve been working on:

*Postgres Support*—in the same week we launched about two years ago,people asked for Postgres support. It brings me pure cathartic joy to finally give this to our community. To be fair, MongoDB has been a perfect solution for our architecture and it’s still recommended. But with a new adapter pattern for databases, you can stand your Payload project up on Postgres and run the same functionality as you can with MongoDB now. The crazy part is that we didn’t compromise on how nesting complex fields works. We could have taken the “easy” road and wrote things to JSON, but we leaned fully into the relational way and built the right tables and native column types for fields all the way throughout.

*Database Migrations*—maintaining a production app while deploying schema changes is something you come to expect from ORMs and backend frameworks, but rarely CMS. Payload 2.0 delivers full, first-party migration support all in TypeScript. We took a lot of care on the developer experience here so that when working with Postgres, thanks to our friends at Drizzle, we generate the migration files in TS that add the tables and fields for you. If you have to manipulate data before or after, you have a clear way forward now.

*Database Transactions*—when a request involves multiple inserts, updates or deletes to the database, you need control to rollback all changes when one part fails. The built-in Payload CRUD operations do this now for you and your custom hooks and other code can too.

*Live Preview*—the ability to quickly draft content and see it in context of a website is a literal game changer. We have taken the best dev experience of any headless CMS and given the editors a reason to demand Payload over the others.

*Lexical Richtext Editor*—our original Slate based editor has seen some great features added, like storing related documents directly in the JSON, uploads and any customizations. Unfortunately Slate leaves a lot to be desired on how to extend it, especially compared to Lexical. In a few short weeks we’ve built up a new editor experience inspired by Medium and Notion. Now type “/” and have embedded relationships, uploads, and custom blocks popping right up to be dropped in. Then drag and drop them to reorder your content. If you still want Slate, we continue to support that too.

We’re not compromising on editor experience. This is how we’re bringing the “head” to the headless CMS.

Building critical applications on top of a CMS may sound like blasphemy but it doesn’t have to be that way.

Thanks for reading! I look forward to hearing what you think.

This is a companion blog post to support the video on a way to integrate Payload CMS SignIn, SignOut and Create Account in a Nuxt JS Application.

There are flat-file CMSes (content management systems) like Grav: https://getgrav.org/

I guess, in some vague/broad sense, config-as-code systems also implement something similar? Maybe even OpenAPI schemas could count to some degree...?

In the old days, the "semantic web" movement was an attempt to make more webpages both human- and machine-readable indefinitely by tagging them with proper schema: https://en.wikipedia.org/wiki/Resource_Description_Framework. Even Google was on board for a while, but I guess it never saw much uptake. As far as I can tell it's basically dead now, both because of non-semantic HTML (everything as a React div), general laziness, and LLMs being able to parse things loosely.

-------------

Side thoughts...

Philosophically, I don't know that capturing raw data alone as files is really sufficient to capture the nuances of any particular experience, or the overall zeitgeist of an era. You can archive Geocities pages, but that doesn't really capture the novelty and indie-ness of that era. Similarly, you can save TikTok videos, but absent the cultural environment that created them (and a faithful recreation of the recommendation algorithm), they wouldn't really show future archaeologists how teenagers today lived.

I worked for a natural history museum for a while, and while we were there, one of the interesting questions (well, to me anyway) was whether our web content was in and of itself worth preserving as a cultural artifact -- both so that future generations can see what exhibits were interesting/apropos for the cultures of our times, but also so they could see how our generation found out about those exhibitions to begin with (who knows what the Web will morph into 50 years later). It wasn't enough to simply save the HTML of our web pages, both because they tie into various other APIs and databases (like zoological collections) and because some were interactive experiences, like games designed to be played with a mouse (before phones were popular), or phone chatbots with some of our specimens. To really capture the experience authentically would've required emulating not just our tech stacks and devices, among other things.

Like for the earlier Geocities example, sure you could just save the old HTML and render it with a modern browser, but that's not the same as something like https://oldweb.today/?browser=ns3-mac#http://geocities.com/ , which emulates the whole OS and browser too. And that still isn't the same as having to sit in front of a tiny CRT and wait minutes for everything to download over a 14.4k modem, only to be interrupted when mom had to make a call.

I guess that's a longwinded of critiquing "file over app": It only makes sense for things that are originally files/documents to begin with. Much of our lives now are not flat docs but "experiences" that take much more thought and effort to archive. If the goal is truly to preserve that posterity, it's not enough to just archive their raw data, but to develop ways to record and later emulate entire experiences, both technological and cultural. It ain't easy!

I took a more traditional approach, focusing on something that's "good enough", which in my case was a cheap VPS and an install of Grav: https://getgrav.org/

Some optional customization for page templates/fonts/CSS, some CI so I can build and deploy it inside of a Docker container, Matomo for analytics that respect privacy (which I already use elsewhere) and some additional web server configuration to hide anything interesting behind an additional login and I'm good. Maybe backups and uptime monitoring if I'm feeling brave, which is what most sites should also have (so copy + paste there).

All of that for under 100 euros per year (could also pay half of that if I didn't host anything else on the server), the blog has actually survived getting on the front page of HN once or twice and requires relatively little maintenance, at least a bit less than a proper install of WordPress, due to its larger surface area.

The best thing is that it's simple enough for me to understand how it works, to be able to move it anywhere as needed and use more or less plain Markdown for writing the blog posts. Here's a quick example of a recent post: https://blog.kronis.dev/articles/ever-wanted-to-read-thousan...

Now all that's left is to find motivation to write more, but at least 90% of my time doesn't go into tinkering with custom fancy solutions, no matter how much I'd love that. Then again, nothing wrong with the alternatives either: 400 euros might be perfectly worth it for some, whereas working with static site generators or even custom CMSes would be a fun experience for others!

I'd check out Grav. https://getgrav.org/

Also, there is a CMS called Grav. Both Gravity and Grav use a very similar (but not identical) font for their logo.

I would use a flat file cms like https://getgrav.org