Hacker News Reader

Comment by weinzierl | Hacker News Reader

I wish there was a browser as secure as Chrome but hackable without restrictions.

I do not think this is a contradiction, at least not from a technical perspective. I am willing to take the responsibility for all actions and modifications I do to my own browser but I need it be secure against all influences out of my control. And I need it come with secure defaults. To be competitive it needs to come without awkward restrictions that e.g. an external sandbox would impose.[1]

I don't think projects like qutebrowser, LuaKit and the many others fit that definition. Not being mainstream means by definition not getting as much security scrutiny as the dominant browsers.

What we really need is a hackable mainstream browser for people that need protection from the bad guys but not from themselves.

[1] I personally would make the concession that supporting a reasonable subset of the web was fair game.

View on HN

The majority of security issues in browsers is in the more low-level components (rendering engine, sandbox, network stack, JS runtime, etc.). None of those small browsers implement any of that themselves, they either build on top of WebKit (e.g. via WebKitGTK, like Luakit), or on top of Chromium (via QtWebEngine like qutebrowser, or via Electron like Vieb).

So you'll mostly need to focus on keeping that up to date. Some distributions (Debian/Ubuntu for example) unfortunately do a bad job at that, but you can also quite easily install them as a binary from upstream.

You still will lag behind a bit on security fixes compared to Chromium directly, that's true. In the case of QtWebEngine, they backport security fixes to the next patch release, and I know of some distributions (I think it was Fedora?) that continuously backport those before Qt releases.

That leaves you with any security issue that's e.g. in the UI, or anything that's in the browser code itself.

For the former, I believe browsers aimed at more technical users can select different tradeoffs that make things more secure (e.g. qutebrowser always shows the punycode-encoded version of a URL if there's non-ASCII in it, while big browsers try to detect whether there are any confusables in it and only show it then - yet new ones are added every once in a while).

For the latter, qutebrowser has had three security bugs in almost 11 years.

Unfortunately I don't share your optimism regarding how hard it is to mess up in the upper layers. I also don't think it really is a technical problem.

The question is if we can get a significant number of eyeballs on such a hackable browser. If enough people have a vested interest, security will follow. Without enough users every effort is futile.

I really hate to be harsh to alternative browsers, because they are all we've got right now now, but three security bugs in 11 years says not much if the user base is that small.

Maybe someone should fork Firefox and make it hackable again. Probably not even that hard. Fix the keybindings, add some first-class-support for userscripting, local extensions and interaction with local system.. All those forks are just focused on adding more privacy or new addons at the moment.

https://firefox.com

Is it as secure as Google Chrome though?

Not hackable enough. Just see the comments here about how extensions like Vimium are limited why projects like the mentioned qutebrowser and Luakit exist.

Also, even if Firefox retains some of its hackability it can be gone tomorrow. I say this with a heavy heart, I personally lost trust in Mozilla as reasonable stewards of Firefox and allies that would protect Firefox users' freedoms. I don't want to drive this thread into that direction, if you're interested there should be enough in my comment history to give you an idea about the reasons and why I think like that.

Reply on news.ycombinator.com