Hacker News Reader

Comment by edpichler | Hacker News Reader

Mozilla really has a big challenge. It is hard to compete with such powerful companies, creating a sustainable business with their mission without succumbing to ads or stopping to protect user privacy. It looks like an unfair game.

I like the Brave business model. I wish to have the same with Firefox.

View on HN

Too bad they built Brave on Chromium instead of Firefox. Which of course given that the company is run by Mozilla co-founder Brendan Eich is a bit of a shame.

Firefox is showing that they are more than technically competent to keep up with Chromium and deliver great performance and functionality. With the work they are doing rewriting Firefox component by component in Rust, Google will have their work cut out keeping up in terms of performance using their C++ implementation. Competition is good.

It’s very easy to build a new browser on top of Chromium/Blink, which is why there are hundreds of Chromium-based browsers out there. In contrast, building on top of Firefox/Gecko, pretty damn hard. Even Firefox devs have admitted this.

All too often I hear sentiments like “too bad they built... on Chromium instead of Firefox”, as if it’s a moral choice. It’s not. It’s an informed technical choice. If you want to change that, you should contribute to Firefox instead of blaming other browser makers.

Somewhat relevant: Mozilla has been working for a while on a project that does supposedly make it relatively easy to build a browser on Android based on their engines: https://github.com/mozilla-mobile/android-components/

This is one aspect that is not often talked about.

If Gecko was easier to build on top of, it would get more adoption in other markets and promote the variety of engines that we sorely need. Mozilla doesn't only win by getting more Firefox market share.

Also not everything is lost. Chromium has a bunch of weird and custom build tooling that also makes it a pain to build on top. Anyone who has tried to package v8 can see that it's mainly catering for Chrome's needs.

>it would get more adoption in other markets and promote the variety of engines that we sorely need.

we don't need variety of engines, we need a non-pofit alternative to google that can throw a punch, which firefox is.

An engine is a technical undertaking (of enormous proportion), and there's essentially no need to have even more of them. If anything it'd just make building websites harder. The issue isn't with blink or gecko it's with Google's control over the browser space. If Firefox was built on blink and had 50% market share I'd be happier than having a hundred engines with 2%.

Strongly disagree with this. When you build on top of Blink you put yourself at Google's mercy: every Chromium knockoff from Opera to Brave or what have you now has to make a critical choice about whether to update their core to the new version that implements Manifest v3, or to keep their base on an insecure outdated version indefinitely.

> With the work they are doing rewriting Firefox component by component in Rust, Google will have their work cut out keeping up in terms of performance using their C++ implementation.

I would expect Blink’s C++ implementation to perform comparably; wouldn’t the main benefits would be security-related?

No, it is very hard to take a complex rendering engine written in C++ and make it multithreaded, secure, and thread safe (i.e. correct).

In theory you could do this in C++ of course but in practice, both Google and Mozilla have struggled with this for many years and both have still some major bottlenecks in their browser rendering pipeline that are effectively still single threaded and can end up blocking the main thread.

This the main reason Rust was invented (by Mozilla) and they've delivered a few component rewrites that leverage e.g. multiple cores. They've been shipping this bit by bit for the last two years (since the quantum release) and they're not done yet. E.g with the latest releases they've started to ship the new web render on some hardware/os combinations.

Faster is a simplistic notion that is hard to quantify but having a more smooth loading page, with less visible stuttering, and higher frame-rates for animation is sort of the main goal here. People also care about memory usage, energy drain on laptops, etc. If you look at all of this then it used to be the case that people went from Firefox to Chrome to improve the experience and you now have people moving the other way.

Mozilla tried to parallelize their CSS engine multiple times in C++, and failed. Rust’s guarantees made it possible to do so. There was a talk about it at a Rust conference called “The Story of Stylo”; I’m on my phone or I’d link you.

Following my curiosity, I found:

The Story of Stylo: Replacing Firefox's CSS engine with Rust

https://www.youtube.com/watch?v=Y6SSTRr2mFU

Woo, 9M lines of C/C++ to 85K lines of Rust..

Author's slides: https://www.joshmatthews.net/rbr17/

No, for the style system is 160K lines of C++ down to 85K lines of Rust. Still a great win!

Those numbers sound much more realistic, ~53% reduction in code size - and you're right, it's a great testament to Rust's conciseness!

Stability also improves when you have less bugs. And you get less memory leaks, what it a big worry for browsers.

But no, I do disagree. Rust is more expressive than C++ and much easier to debug. I expect the new Gecko development to be easier and faster than Chromium, what leads to more features.

> Too bad they built Brave on Chromium instead of Firefox

Weren't the initial versions of Brave based on Firefox? I rem reading it somewhere that they later had to move to Chromium for one reason or the other.

Brendan Eich:

> No real choice on mobile. On desktop we started in 2015 w/ Gecko, did full evaluation against chromium, latter won. No benefit in going back to Gecko (Firefox faces diff tradeoff but on back foot now breaking XUL addons). WebKit lineage + Chrome market power => de facto standard.

https://twitter.com/BrendanEich/status/950209816902774785

All publicly released versions are Chromium-based.

https://www.zdnet.com/article/brave-new-browser-eich-returns...

> Firefox is showing that they are more than technically competent to keep up with Chromium and deliver great performance and functionality.

Firefox's performance on android is atrocious.

They've reinplented their browser recently. Still in preview with extensions support in development. But it's faster than Chrome in a lot of cases.

https://play.google.com/store/apps/details?id=org.mozilla.fe...

Nobody would fork Firefox today, as they are just following Google reimplementing Chromium and as a consequence always behind and catching up, just burning resources. While they could fork Chromium themselves and focus instead on features that create an advantage and attract users.

While I agree that a lot of features from Chrome appear later in Firefox, I don't think it's true for the engine.

As an example, asm.js, the technology which became Web Assembly has been invented by Mozilla folks.

Firefox seems reluctant to innovate in the web though. You need to at least push for some change in order to see it. They 've barely wet their feet with that, very reluctantly and not with enthusiasm (like with Persona).

Reply on news.ycombinator.com