Many reasons, in no particular order:

- It is extremely slow and resource intensive. Opening any link/page takes at least 3 seconds on my fastest computer, but the content is mostly text with images.

- It cannot be used without JS (it used to be at least readable, now only the issue description loads). I want the bug tracker to be readable from Dillo itself. There are several CLI options but those are no better than just storing the issues as files and using my editor.

- It forces users to create an account to interact and it doesn't interoperate with other forges. It is a walled garden owned by a for-profit corporation.

- You need an Internet connection to use it and a good one. Loading the main page of the dillo repo requires 3 MiB of traffic (compressed) This is more than twice the size of a release of Dillo (we use a floppy disk as limit). Loading our index of all opened issues downloads 7.6 KiB (compressed).

- Replying by email mangles the content (there is no Markdown?).

- I cannot add (built-in) dependencies across issues.

I'll probably write some post with more details when we finally consider the migration complete.

I’m imagining a comfy bevel on the bench outside your hostile where your boarders can sit happily.

"High performance has shifted into how fast we can serve bloat."

If remove ads and behavioural tracking, speed is faster

But goal of Big Tech, who make the popular browsers, is to make speed faster (fast enough) _with_ ads and tracking

User wants fast speed. User does not want ads and tracking. Big Tech wants users in order to target with ads and tracking. Big Tech tries top deliver fast speed to keep users interested

User can achieve fast speed _without_ ads and tracking

I do it every day. I do not use a large propular browser to make HTTP requests nor to read HTML

More memory means more memory for my website to take up!

They're like 10x more complex and you don't need most of their functionality for just a frontend.

That said I wish there was something a little better than cgit

My guess is gitea and forgejo don't render well in Dillo.

I remember installing Debian on my OLPC XO-1. Dillo and Netsurf were the only browsers that I even tried running on that thing (w. 512MB RAM). Netsurf had better compatibility, but Dillo was noticeably faster and more responsive. Truly a pleasure to use when it supported the site I was on.

And dillo still works great on NetBSD :)

I think it is becoming more important to i386 BSD, especially since i386 OpenBSD can no longer build Firefox, Seamonkey and IIRC Chrome on i386 systems.

I have been using dillo more and more as time goes on, plus you can get plugins for Gemini Protocol and Gopher.

Yes, thanks for the reminder. This is what they write about:

https://dillo.org/post-sitemap.xml

At some point I should investigate if we can fill a complaint to get it taken down at least. Here is more info: https://dillo-browser.org/dillo.org.html

I wrote my own:

https://git.dillo-browser.org/buggy/

It fetches the issues from GitHub and stores them in <number>/index.md in Markdown format, with some special headers. I then keep the issues in a git repository:

https://git.dillo-browser.org/bugtracker/

So we have a very robust storage that we can move around and also allows me to work offline. When I want to push changes, I just push them via git, then buggy(1) runs in the server via a web hook. This also tracks the edit changes.

While typing, I often use `find . -name '*.md' | entr make` which regenerates the issues that have changed into HTML as well as the index, then sends a SIGUSR1 to dillo, which reloads the issue page.

The nice thing of allowing arbitrary HTML inline is that I can write the reproducers in the issue itself:

https://git.dillo-browser.org/bugtracker/tree/501/index.md#n...

Closing an issue is just changing the header "State: open" by "State: closed", often with a comment pointing to the merged commit.

https://git.dillo-browser.org/buggy/

Not really. There was this list but it is outdated: https://dillo-browser.org/old/css_compat/index.html

Probably the best indicator of which features are supported is to pass as many tests as possible from WPT that cover that feature.

I did some experiments to pass some tests from WPT, but many of them require JS to perform the check (I was also reading how you do it in blitz). It would probably be the best way forward, so it indicates what is actually supported.

Check out: /etc/dillo/dillorc

There are options here for (my setup below):

geometry=1600x900

increase font_factor=1.75

bg_color=0xFAF9F6

Start and Home pages too.

DIllo is much lighter and it supports Gopher, Gemini, Info, Man and potentially in a further future, URL rewritting plugins.

Yes, we've deliberately tried to make use of existing libraries (either from other browser engines or general purpose libraries) where possible.

The main thing we pull in from Servo (which is also shared with Firefox) is the Stylo CSS engine, which is a really fantastic piece of software (although underdocumented - a situtation I am trying to improve). I was recently able to add support for CSS transitions and animations to Blitz in ~2 days because Stylo supports most of the functionality out of the box.

(the other smaller thing we use from servo is html5ever: the html5/xhtml parser)

We also rely on large parts of the general Rust crate ecosystem: wgpu/vello for rendering (although we now have an abstraction and have added additional Skia and CPU backends), winit for windowing/input, reqwest/hyper/tokio for HTTP, icu4x for unicode, fontations/harfrust for low-level font operations, etc.

And while we're building the layout engine ourselves, we're building it as two independently usable libraries: Taffy for box-level layout (development mostly driven by Blitz but used by Zed and Bevy amongst others), and Parley for text/inline-level layout (a joint effort with the Linebender open source collective).

---

I wish that the Servo project of 2025 was interested in making more of their components available as independent libraries though. Almost all of the independently usable libraries were split out when it was still a Mozilla project. And efforts I have made to try and get Servo developers to collaborate on shared layout/text/font modules have been in vain.

Took me a sec to understand you meant comma as “decimal point” :)

Yes. It currently compiles to WASM but doesn't run. But that's just a matter of plumbing it in properly.

If you were running it "headless" (which is supported), then it would probably work today.

There would also be the option of using Taffy and/or Parley (the layout libraries) without the rest of Blitz.

That's a bit of an open question at the moment. The obvious choice from a Rust ecosystem perspective (easiest to integrate) would be Boa (https://boajs.dev/). It has excellent standards conformance, but terrible performance. We'd need to test to what extent that would be an issue in practice.

Other engines on my radar: quickjs, hermes, primjs (and of course V8/JSC/SM, but they don't exactly fit with the "lightweight ethos").

There is also the possibility of bindings with more than one engine and/or using some kind of abstraction like NAPI or JSI.

I consider Netsurf to be a beautiful and excellently well-done browser in its own right, so you can't go wrong with either.

But by comparison, Dillo is much more lightweight than even Netsurf (!!), much more brutalist, and a bit more idiosyncratic and the kind of texture and feel of how tabs behave, how you handle bookmarks, how you do searches. Dillo uses fltk while netsurf uses gtk3, and a lot of the resource usage savings in differences in vibe and feel come from that by itself. Netsurf is much more familiar if your baseline is standard modern browsers, and Netsurf does a better job of respecting CSS and rendering pages the way they should look.

Dillo can take a little bit of getting used to but it's a usable text oriented browser that I think is probably as good as it can possibly get in terms of minimalist use of resources, although at a rather significant compromise in not rendering many web pages accurately or using JavaScript or having an interface intuitive to the average person.

Huh, somehow I had the idea that Raph Levien wrote it. The PNG here says corvid, Jeremy Henty, Johannes Hofmann, Jorge Arellano Cid, Rodrigo Arias Mallo (which is presumably you), Sebastian Geerken, and "other".

It's good to see that it's active again!

Yeah, I tried to reach out to Google back when they introduced the JS-wall, but they seem to have an AI chatbot acting as a filter, so I didn't spent much energy.

Later they also blocked other non-JS browsers like links or w3m, so I assume they no longer care. They used to maintain several frontends that worked in really old devices.

I don't think there is any User-Agent that works today, however you can still use the Google index via other search indexes that can fetch Google results without JS (for example Startpage still works). However, it is probably a good idea to have more options available that have their own independent index engine (for example Mojeek). Seirdy has a very good list: https://seirdy.one/posts/2021/03/10/search-engines-with-own-...

Only several weeks ago was when they broke it for all UAs I could try. If anyone has figured out one that still works, please do tell.

Incidentally, DDG still works without JS.

On the same machine, Google works fine in links, but fails in dillo, seamonkey, firefox without js.

Yes, I have been using

https://lite.duckduckgo.com/lite

for searching in Dillo. Once in a while you may get a captcha from DDG that is far better that any other captcha I have ever seen. The captcha is easy to use and can be a bit fun :)

Nope, it uses https://www.fltk.org/