I mean, web browsers are pretty fucked design-wise these days and Edge sucks, but the crawl towards only a single web rendering engine that's Google controlled is pretty bad

@cwebber I agree. We still have Firefox and some minor engines. And Firefox steering headless towards Rust is another can of problems.

But can we say Edge went over the edge?
@cwebber apparently this was the origin. But anonymous sources are good, and bad at the same time.

So far I would call it really rumours.

@ng0 Hm, I didn't realize it was only rumors for now. Hopefully it's just rumors!

@kelly_clowers I suspect @ng0 is talking about packaging difficulties since Rust is self-bootstrapping and that creates the usual difficulty for reproducible operating systems where you have to bootstrap the compiler with an older version of itself. I could be mistaken though, I shouldn't speak for ng0 :)

@ng0 @cwebber @kelly_clowers er, ok, but how is that any different than C++ which blink and gecko are written in? Just because it changes less?

@kelly_clowers @cwebber cwebber, you might want to summarize our efforts and findings. I'm a bit preoccupied. Just this, it is different.

@kelly_clowers @ng0 since "We Have To Do It" there's already work being done on C/C++ compilers to be fully reproducible, see Mes

It's just annoying as a packager to deal with these things mostly because language authors are mostly "happy" that the umbellical cord has been cut when self-compiling is possible, but if you want to avoid "Thompson Attacks" ie "Reflections on Trusting Trust" it's more work if a version pre-self-hosting isn't maintained

@kelly_clowers @ng0 Don't get me wrong, from a security perspective the move to a memory safe language is a *huge win* :)

@cwebber preach ✊🏽 I'm not sure what the state of Chromium is these days, but Chrome is a bloated buggy mess. Regardless of reasons, data parasite Google needs less influence in our lives.

@cwebber its nigh impossible to find a decent web browser that isn't chromium/blink based or firefox

@pupy @cwebber well, blink and gecko are the only 2 really solid engines these days, with WebKit and Servo behind, falling and rising respectively. A good browser based on any engine is pretty well non-existent though. SeaMonkey is falling behind on gecko version, and has no good path forward, and FF is FF, so. And Chrome is trash

@kelly_clowers @cwebber could do with just tossing all of it in the bin and starting over

@pupy @kelly_clowers extremely, extremely hard given the way web standards moved to "it's whatever the browsers are implementing" and the sheer number of technologies users expect these days

expect about a 30 million dollar investment today to get that to happen with a comparable number of "features"

@cwebber @pupy @kelly_clowers

Well if WebAsm progress far enough,we could hypothetically write a browser (the HTML,CSS and JS parts) in it and reduce the compiled part to just a WebAsm interpreter,no bigger than JDK


@cwebber @kelly_clowers well yeah I didn't mean it seriously, was more a facetious remark than anything else haha

web tech is fucked

@cwebber @kelly_clowers what was that post I saw the other day that was like

8 years ago we had small portable computers specifically for browsing the web and nowadays its one of the most resource intensive tasks you could possibly do lmao

@cwebber @pupy @kelly_clowers i'm pretty far from expertise, but that almost feels low - i mean, what do you need to pay in annual salaries and infrastructure expenses just to _keep up_ with changes once you somehow get to feature parity?

this is compounded by the fact that there's mostly no such thing as graceful degradation or fault tolerance on the web at this point. an engine with like 90% of the featureset still wouldn't _work_.

@cwebber @kelly_clowers @pupy This discussion actually makes me start thinking that the Web should be simplified: it's really the sheer amount of different stuff people expect from a Web browser which keeps reducing the number of rendering engines further and further. I suppose it should be possible to provide an similar amount of functionality with better written (= composable!) standards.

@scolobb @cwebber @kelly_clowers @pupy I remember when the web was a document display system, not an application platform…

@kelly_clowers @pupy @cwebber Judging by the tendencies in GUI development before Web 2.0, I think people would have converged to something HTML/CSS/JS-like anyway (I'm thinking about describing interfaces in XML, for example). But it needn't be so contrived and full of cruft.

Also, thinking about the discussion in the second episode of Libre Lounge, I'd generalise what @cwebber says about NPM vs. other package/ecosystem managers: a lot of Web-related standards have probably been developed with too much focus on solving a single practical problem (=> poor composability => a lot of boilerplate and cruft).

Given that we have TBL in the conversation now about the way that the web as an application platform is detrimental to accessibility, what's the chance of us getting a low-cruft content model that leaves the client in charge of presentation and behavior?
@pupy @kelly_clowers

@cwebber @grainloom I feel like this is an expected result of the sheer complexity of modern web standards. By making them this complex, they've basically ensured that there will be very little competition, because the cost to compete is too high.

@alva @grainloom yeah but I think it wasn't an intentional or malicious anti-competitive move, but rather emergent behavior

@cwebber honestly we had 3x more problems with Safari than Edge. They don't want their own CSS stuff all the time at least.

@cwebber @ink_slinger Yeah but if everything uses Blink/WebKit then Konqueror will have been vindicated post-mortem!

Man, I still miss Konqueror. That might be the single biggest reason why I switched over primarily to Linux for my desktop OS lo so many years ago . . .

Sign in to participate in the conversation

Octodon is a nice general purpose instance. more