Live data from Hacker News

WebAssembly: Mozilla Won

robert.ocallahan.org

371–375 of 375 posts

Re: WebAssembly: Mozilla Won

#371
post #369
post #367

Earlier quoted context omitted.

Moving goal post? Preloading is useful on every single web page that, for example, loads a script. Not just 3D games, but any web page with scripting.

Keep reading the entire comment: the point wasn't that rel=preload isn't useful but rather questioning the assertion that it's this huge game-changing feature which is causing ordinary people to switch browsers. If you actually run benchmarks, it's nice but hardly a game-changer, especially in the post-HTTP/2 world. If you're concerned about cold page load times most sites will see significantly greater benefit from…

I don't recall saying that it was "game-changing", I said that it was useful for a large number of web pages, unlike webassembly which is useful for a very small number of specialized needs.

Same for modules, or Intersection Observer or many other features that FF is not shipping. Stop trying to prove me wrong and understand my point. Firefox is slow on stuff that makes regular old web developers lives easier, because they're chasing 3d games.

Re: WebAssembly: Mozilla Won

#372

Earlier quoted context omitted.

I hope they never go with process-per-tab in Firefox. It's one of the things I dislike about Chrome.

I thought that's what Electrolysis was. We've switched to an ESR release with e10s disabled because of an incompatible add-on listed as compatible. Annoying, because the idea is if an add-on doesn't work with e10s, e10s will be automatically disabled. And of course in this case it doesn't since the add-on works "great" with it!

I believe the main thing Electrolysis does is split the UI and rendering into separate processes. It also creates separate processes for some other tasks. But as far as I know it doesn't give each tab a separate process.

Re: WebAssembly: Mozilla Won

#373
post #224

Earlier quoted context omitted.

I donate to Mozilla every year, and I use Firefox on phone and desktop. I like that I can install ublock origin on mobile firefox. Chrome does not support this option.

You pretty much sum things up as they were for me half a year ago. I'm not happy about them putting the sword of Damocles over Vimperator and Pentadactyl though, which is why I'm mainly using Qutebrowser on the desktop now. That said, Mozilla is still awesome and FF will most likely remain my second go-to browser.

One other thing I stumbled upon, the responsive design mode in Chrome was not working correctly. I was pulling my hair out when developing with it. I switched over to Firefox in their responsive mode and it worked correctly and reflected what I saw on an actual phone.

What has always frustrated me most about Chrome is that certain bugs or fixes can take an eternity to fix despite there be many reports on an issue.

Re: WebAssembly: Mozilla Won

#374
post #371
post #369

Earlier quoted context omitted.

Keep reading the entire comment: the point wasn't that rel=preload isn't useful but rather questioning the assertion that it's this huge game-changing feature which is causing ordinary people to switch browsers. If you actually run benchmarks, it's nice but hardly a game-changer, especially in the post-HTTP/2 world. If you're concerned about cold page load times most sites will see significantly greater benefit from…

I don't recall saying that it was "game-changing", I said that it was useful for a large number of web pages, unlike webassembly which is useful for a very small number of specialized needs. Same for modules, or Intersection Observer or many other features that FF is not shipping. Stop trying to prove me wrong and understand my point. Firefox is slow on stuff that makes regular old web developers lives easier, becaus…

The problem is that you made a very broad and sweeping claim and then picked a very weak supporting point for it. Web Assembly may not change the game for every or even most sites but for sites which are performance sensitive it's an extremely user-visible improvement.

> Stop trying to prove me wrong and understand my point.

I think you should focus on making your point more clearly rather than defending what was clearly a. For example, you cite modules as something which is apparently a big deal for web developers but not shipped by Firefox. Sounds like Mozilla needs to get cracking … unless you know that only Safari has shipped it and the Chrome, Firefox, and Edge teams all have it available but behind a feature flag for testing:

https://www.chromestatus.com/feature/5365692190687232

That doesn't support your narrative that the Firefox team is ignoring this or that they're behind the market — and since anyone who isn't targeting only the latest version of Safari is either polyfilling or continuing to use their existing strategy, so there's an upper bound on how bad that can be, too.

Similarly, with Intersection Observer you can see that it was enabled in FF50 but had stability issues which lead to it being disabled and is likely to be re-enabled in FF54 based on testing. Unless you have some evidence that the developers who were working on that were pulled off to work on WebAssembly it doesn't seem like an especially compelling argument.

Again, I'm not saying that any of these things are useless — only that the narrative you're insisting on where Mozilla is ignoring web developers doesn't seem to be well supported by the evidence. At least for the projects I work on, I'd level that criticism at Safari or Edge first and in the much fewer cases where either Chrome or Firefox has a bug or limitation I need to work around it's about as likely to be Chrome as Firefox requiring extra work.

Re: WebAssembly: Mozilla Won

#375
post #202

WebAssembly and PNaCL always seemed like a hack. A very elegant thought-out hack, but still a hack. It required a 64 bit OS but could only ever operate in 32 bit space. Should have been dead on arrival because of that fact alone. For something that was trying to bring C/C++ and video games into a browser setting a 4GB max on memory should have been a non-starter.

I'm afraid pretty much every part in the pipeline of getting websites to your computer and shown on screen is a big giant hack of which it's a wonder that it even works at all, most of the time.

HAHA, you're correct
Post reply on HN