The Weaver web browser

25 May 2026

I think the world needs more web browsers, so I'm making one. It's an experiment - I want to see if you can take a really simple approach to rendering pages. To set your expectations, this is a browser for the "document-based" web - no javascript, and so no web apps.

The idea

We all know the web is in a bit of a state right now. As it has morphed into a web app delivery platform, its complexity has ballooned. Pages have become bloated. If you want to make a new browser that is capable of rendering the modern web as it is, it's an enormous undertaking. This is a bit of a shame for those of us who still see the promise of the web as a set of rich, interlinked documents.

There are simple browsers out there, like Dillo and netsurf. Dillo has been around since 1999 and development restarted a few years ago. It's great fun, but I find that I'm always having to toggle the embedded and remote CSS options to get the page in a usable state. It tries to render pages as best it can, but problems occur because its implementation of CSS is only partial. netsurf has a similar problem on many sites.

Weaver will have more in common with text-based browsers like links. These feel like true "user agents" in the sense that they largely ignore the intent of the web designer when it comes to presentation, and instead render the page in a way that the user has asked for, i.e. just text. I want to see if we can take this idea and extend it to graphical browsers. We can have rich text and images and tables and forms, but they'll be laid out like a document. Maybe we can detect structures like a tree of comments on a discussion board, or a forum thread, or a list of news headlines, and show them in a nice way without having to worry about styling too much.

Some more vague ideas

• If you use Dillo or a text-based browser, on big sites like wikipedia or news sites, you'll first be met with a wall of navigation links at the top of the page. There might be a "skip to content" link for accessibility, but why not just move all of these links out into a navigation sidebar? This could also contain a table of contents for the page based on its structure.
• We can make pages a bit richer than they are in normal browsers. For example, tables should be sortable and filterable! Wikipedia implements sorting with javascript, but it should just be part of the browser. Image alt text should be viewable (not just the title attribute).
• Page metadata should be viewable by the user, such as links to RSS feeds, and Open Graph data. If it's in the HTML source, the user might want to know about it.
• We can support cookies for logins, but the user must have full visibility and control over this
• I want the browser to have as small a footprint as possible. I want to do this to help keep old hardware alive, but also new hardware that's not very powerful. In fact I'd love for this to be able to run on a (beefy) microcontroller, say an ESP32 with 8MB of RAM. (I like microcontrollers ok)

Weaver is on codeberg. It can't even follow links yet so don't get too excited.

Do you have any comments or suggestions? Please get in touch via email or Mastodon! Or just let me know if you think this is interesting.