|
If I understand https://arewefenixyet.com/ correctly, being on Norway just landed me in the early Fenix (Firefox on Geckoview on Android) rollout. I had no idea it was coming, and uh I'm not sold on this being finished to the point where they can just quietly move people over and expect everything to go smoothly. I've had multiple UI freezes (of the close and relaunch Firefox kind), I miss a few add-ons, the add-on author (a friend of mine) has been looking at it but doesn't even understand where to begin debugging why they are "not yet supported", I'm logged out of everything (and have to accept a thousand cookie warnings again), I'm not super sold on the UI changes, and the performance seems better in some places but worse in others; definitely not a clear win. Oh, and someone mentioned about :config and view-source:// don't work right now, and I miss my Top Sites list. I mean, I understand that this is the right way to move, and in the long run it absolutely for the best - but this feels like they didn't want to bother keeping the old version alive and kicked a beta over the fence.
|
#
¿
Jul 29, 2020 01:18
|
|
|
|
| # ¿ Apr 29, 2024 10:15 |
|
|
Yeah, I spent last evening reading bug reports and discussions. They've been incredibly noncommittal so far. (I also miss holding back for a history list, but that seems to exist in the nightlies now?)
|
|
#
¿
Jul 29, 2020 17:11
|
|
|
At least Ctrl-Shift-T makes some sort of sense as a modification of Ctrl-T.
|
|
#
¿
Aug 12, 2020 20:59
|
|
|
Huh, a mere three weeks after they started rolling it out, Mozilla have officially published a real announcement and a new codename for Fenix - it's now "Firefox Daylight". As one of the first to get it, I have a distinct "we only really care about users in the big markets" - feeling.
|
|
#
¿
Aug 25, 2020 17:24
|
|
|
The official explanation for about:config is that there are multiple known ways to break Fenix so badly with it that you need to wipe and reinstall to recover, and they don't want that sort of foot-shooting available in the stable browser. The secondary argument is that about:config does less now, since it only configures Geckoview, not the (android-native) UI code - unlike in fennec where gecko drew everything. It's enabled in beta and nightly, though. (They also claim they want to add more settings in the UI to cover most of the uses people had for a:c.)
|
|
#
¿
Aug 30, 2020 04:03
|
|
|
Ihmemies posted:They booted all their devs Not really, no. The Gecko/Firefox teams themselves are untouched, though they did fire people working on the devtools, some internal tooling, and some qa/security (after merging a few overlapping groups). Not great, but very different. They did fire the entire "new browser engine in rust" group, whatever that was called - but the grumblings I saw on twitter suggests that those people were kind of detached and didn't actually deliver that much of practical use; the rust parts that are actually in Firefox took a fair bit of reworking from the normal developers first.
|
|
#
¿
Aug 30, 2020 14:34
|
|
|
Having their own rendering engine is one of their main reasons for existing - it gives them some clout in designing web standards. Without it, we would be down to Google defining the web as "whatever Google programmers decided to do with Chrome this week". As is, they do at least have to pay some lip service to compatibility and open standards forums. (Ok, Apple can also take WebKit in their own direction, though I don't know how far they want it to diverge.)
|
|
#
¿
Sep 10, 2020 10:06
|
|
|
At least from my perspective, shaping the future of the web and making sure Google doesn't have a tech monopoly is their reason to exist, and the browser wrapped around their tech is just the means to get there. Giving up the engine to make a better browser would be as counterproductive as green party giving up the environmental policies to get more voters - or peeing their pants to stay warm, as my dad would say.
|
|
#
¿
Sep 10, 2020 23:04
|
|
|
dougdrums posted:having an existential crisis over a web browser itt It's kind of like when amazon or walmart takes over - you lose something when there is only one real alternative, even if it works well enough.
|
|
#
¿
Sep 28, 2020 08:40
|
|
|
Fame Douglas posted:Yeah, wouldn't want to lose the worst rendering engine on the market, what a terrible outcome that would be! In a contest with two entrants that doesn't really say much.
|
|
#
¿
Sep 30, 2020 15:01
|
|
|
Webfonts: Just say no.
|
|
#
¿
Oct 23, 2020 16:55
|
|
|
There are also the rare cases where the "local" server at the exit node is faster than your actually local one - github in europe springs to mind. Something about data shaping on 4g mobile networks also apparently responds better to the single stream of vpn data, but I suspect that to be mostly theoretical. But yeah as a rule of thumb, sending all your traffic via a third party will make things slower. The best you can hope for is for it to not be noticeable.
|
|
#
¿
Oct 31, 2020 05:15
|
|
|
Both vertical tabs and tab grouping do sound like potentially nice features to support natively, mind.
|
|
#
¿
Jan 3, 2021 18:40
|
|
|
If you're still using Office 2010 like we do at work, I suggest going into the compatibility settings on the shortcut you use to start it, and setting the scaling to "managed by application" - that seems to both scale reasonably well and remain sharp.
|
|
#
¿
Mar 22, 2021 14:16
|
|
|
I guess that if the threads are lightweight enough it's edges into just being an implementation detail, sort of like using hundreds and hundreds of cheap threads to sleep waiting for traffic on incoming sockets in a server. It's definitely going to stress test some parts of the OSes it runs on, though.
|
|
#
¿
Mar 23, 2021 22:14
|
|
|
That is incredibly annoying, yeah. Restarting and restoring tabs usually fixes it, but shouldn't be needed. I do think it's a straight up bug, though.
|
|
#
¿
Apr 1, 2021 03:17
|
|
|
"Blocking webRequest" is an API for filtering every single request to any URL. Basically, an extension can register itself as interested, and then whenever anything opens any URL, FF calls a function in the extension with the URL as a parameter, and waits for it to return. This is incredibly powerful, since the addon can block anything it doesn't like [1], rewrite any request [2], or completely take over requests matching specific patterns [3]. It's got some glaring downsides, though. First of all, this is an incredibly powerful position to be in: You really, really have to trust any code that gets to do this sort of total control. It also has some performance downsides. The big one is the "blocking" part: Every single request has to wait for the extension function(s) to return before they can continue, which can add up. It's also apparently a bit messy in what happens when multiple extensions have differing opinions on an URL; as I understand it FF resolves conflicts by just stopping the other extensions as soon as the fastest one with an opinion has returned. A while ago, Chrome added a completely different API for adblockers ("declarativeNetRequest"). It's simple enough: Extensions can register a list of simple patterns to allow or block, and then chrome itself does the filtering. This resolves the speed and security issues, but doesn't allow the more interesting weirdness from webRequest - and some people think the limits on block list length are too low. Firefox and Chrome both support a general standard for how to write extensions ("Manifest"; they're phasing in version 3). This is nice - it means a number of extensions can easily work on every desktop browser. However, Chrome/Google has decided to turn off webRequest, while Firefox does currently not support declarativeNetRequest. This announcement is Mozilla saying they'll add DNR without removing webRequest ("until there’s a better solution which covers all use cases we consider important "), which allows both existing FF extensions and unmodified chrome ones to work. Which is nice - there was a reasonable worry that they'd just do like Google and replace one with the other. [1] This is how adblockers traditionally work: They have a configurable list of patterns to block, and try matching every URL against every pattern to determine if it gets through [2] Useful for extensions like Https Everywhere - just rewrite http: urls to https: unless there's a reason not to [3] Which can be used for things like adding support for gopher:// or redirecting certain URLs to external apps
|
|
#
¿
May 28, 2021 01:06
|
|
|
This is very much a "multiple things can be true at once" sort of situation. Is it possible for addons using webrequest to slow things down? Unquestionably. Does ublock origin actually make requests slower for the typical user? Not in any meaningful way. Is google in a position of near absolute power over everything you do in Chrome? Of course, and that is a bit worrying. Can an addon that has suddenly been taken over by Moldovan hackers use webrequest to do a lot more harm than without it? Yeah, probably?
|
|
#
¿
May 28, 2021 17:36
|
|
|
Oh chrome is absolutely out to mess with adblockers, and not to be trusted here. That does not mean that webrequest couldn't be improved, just that declarativeNetRequest is not the right answer.
|
|
#
¿
May 28, 2021 20:09
|
|
|
I wonder if the non-google chromium browsers are enough of a critical mass to keep a fork going? I mean, both Opera and MS used to develop their own entire engines; so in theory it should be doable. Of course, opera fired all those developers to save money shortly after switching, and I imagine MS repurposed theirs as well. Still.
|
|
#
¿
May 28, 2021 21:22
|
|
|
Safari doesn't really use chromium, though. The family tree goes something like this: KDE develops KHTML to power their konqueror web browser. Apple grabs a copy and starts adding their own things to it, sporadically throwing a big blob of undocumented changes back to KDE. They eventually part ways. Apple renames their version WebKit - and for GPL license reasons, it remains open-source. Google needs a rendering engine for their new browser, and decides on WebKit. For a few years, google and apple manage to cooperate on WebKit - though each has their own javascript engine bolted to it. Eventually, they, too, disagreed over ... something (presumably the same "throw it over the fence" approach apple had with KDE), and from there on they have been independent trees with a common ancestor. Google's fork changed its name to Blink shortly after this happened. Chromium is Blink + V8, while Safari is WebKit 2 + JavaScriptCore (or whatever they call it these days). As for Firefox, my impression is that it has been hard to pull out just the rendering engine, which is why the projects you do see (like Tor browser) are more like Firefox with skins and extensions. At least on the Android side, they have worked on making a drop-in replacement for the chromium-based HTML view you can use in apps, which probably means some degree of modularization - but that's a recent development. Computer viking fucked around with this message at 21:51 on May 31, 2021 |
|
#
¿
May 31, 2021 21:44
|
|
|
Oh that's kind of neat.
|
|
#
¿
Jun 1, 2021 13:04
|
|
|
That looks like a broken webfont, but I can't quite imagine how that would happen.
|
|
#
¿
Jun 2, 2021 22:15
|
|
|
GreatGreen posted:Jesus almighty loving christ why does every firefox update make it more and more obnoxious. Tabs are even bigger now, and all colored the same but with a thin box outline around the active one? I may be weird for using their third color scheme - but I find the active tab to be fairly clearly indicated? 
|
|
#
¿
Jun 4, 2021 23:27
|
|
|
GreatGreen posted:I just had the default scheme and what I saw was a grey tab on a very slightly differently colored grey background. Oh yeah, that's awful.
|
|
#
¿
Jun 5, 2021 02:22
|
|
|
I'll say that the account/sync system is quite useful - I don't use it every day, but being able to e.g. quickly see what tabs I have open on my phone and opening a copy on my desktop is neat.
|
|
#
¿
Jun 13, 2021 15:23
|
|
|
Serephina posted:For gently caress's sake, really? The article even lists why it's an annoying change: No, I don't need to clutter my DL folder with small disposable files like pdfs and torrents. OTOH, I'm all for this - accidentally clicking open and then wondering why the file I had to go through two logins and a captcha to get to is nowhere to be found was a bit of an annoyance.
|
|
#
¿
Aug 7, 2021 17:13
|
|
|
It's actually kind of convenient on the work network, where clicking any of the many mailto links opens Outlook and that is genuinely what I wanted to happen. Outside work, getting a Gmail tab when I click a mail address isn't useful very often, but ... it's not wrong, either? I can't find it in me to be annoyed at it.
|
|
#
¿
Sep 20, 2021 21:20
|
|
|
Click and hold to count upwards; wait for the number to roll over if you want to go back.
|
|
#
¿
Nov 4, 2021 23:30
|
|
|
It's probably worth being precise here: they try really hard to make links opened from most new Microsoft apps open in Edge. This includes Start menu searches, help links, News and Interests, Mail, that phone link thing, the widgets, and probably a bunch more - but not things opened from any third party software (or Office). Of course, that's still quite bad. The original article leans even harder into the "this has got to be some sort of antitrust violation" angle. (Daniel's an old Opera employee, so I suspect he's slightly bitter.)
|
|
#
¿
Nov 16, 2021 00:50
|
|
|
If you haven't specifically set anything, I think it follows your OS settings now?
|
|
#
¿
Dec 9, 2021 02:47
|
|
|
It could also be a service worker, since I'm sure twitter does all sorts of modern web app and caching work in the background.
|
|
#
¿
Dec 13, 2021 17:21
|
|
|
It also kind of sounds like the revocation server for a registrar is misbehaving. I think firefox handles that differently than the Chromium family, and it would explain how it affects a seemingly random but consistent selection of sites? On the other hand, gmail and youtube are both signed by Google Trust Services (same root and intermediate certificate, even), so if they behave differently it's probably not that.
|
|
#
¿
Jan 11, 2022 16:47
|
|
|
That sounds ... reasonable? It's a niche problem, but one very few companies are in a position to handle. Of course, if the participants are limited to firefox users there may be some problems with trying to generalise your findings to all web users.
|
|
#
¿
Feb 3, 2022 19:37
|
|
|
WattsvilleBlues posted:Aside from the fact I use uBlock Origin on any browser because I'm not a loving animal, I don't think I've ever deliberately clicked on an ad. In my recent mucking around with Linux distros, for the craic I sometimes go to random sites before I install UO and some of them are absolutely unusable. On one hand, you're entirely right; large swathes of the Web are horrible and nearly unusable messes without an adblocker. On the other hand, I'm kind of uncomfortable with how the adblocker teams are now the gatekeepers that decide if a site without other income will be viable or not. My partner is trying to make a living writing a tech site, and his adblock ratio is something like 75%. That's kind of to be expected; he jokes that he should have picked a subject with a less tech savvy audience. What's worse is how he has an ongoing struggle with the major ad blockers to keep a non-intrusive "I see you're using an adblocker; consider disabling it so I get paid" text box and a small donation button (underneath the article) unblocked - it's like they're trying to make sure he can't earn anything. e: Not that this is a new problem - and I don't have any good suggestions for a solution. Computer viking fucked around with this message at 17:58 on Feb 14, 2022 |
|
#
¿
Feb 14, 2022 17:54
|
|
|
~Coxy posted:You've probably thought of this but I think the only good solution for this is to build a Patreon following. Yeah. The downside is that if the patreons get any benefits, he's suddenly selling goods or services from a European country and needs to start thinking about VAT. IIRC Patreon can help for certain categories but not others, and there's a new European one-stop VAT solution that may help, and there's special considerations when selling to people in your own country. It's probably not too bad in practice, but it feels like one of those things you should have a plan for before you start.
|
|
#
¿
Feb 15, 2022 11:21
|
|
|
Yup. The stop/reload button in the menu is iffy, too- though I have the address bar at the bottom, FWIW.
|
|
#
¿
Feb 17, 2022 22:45
|
|
|
astral posted:Something exciting on the way? Just the rollback of the "the URL bar is focused in a new window with a homepage" change.
|
|
#
¿
Sep 25, 2022 13:40
|
|
|
Geemer posted:That was in the freshly released .1 version. Oh, right.
|
|
#
¿
Sep 25, 2022 13:46
|
|
|
|
| # ¿ Apr 29, 2024 10:15 |
|
|
KozmoNaut posted:Apparently Firefox on Android still doesn't properly support webauthn etc., so trying to use a Yubikey or whatever forces the use of Chrome. Here in Denmark, everywhere you need to authenticate in an official fashion is using a 2FA solution called MitID. I suspect this is closely related to Norwegian BankID - and I'm very happy I never moved on from the key generator; my last one survived a full decade on my keychain (before it took a 60C clothes wash ... and it still worked afterwards, but I ordered a new one just in case). Most apps cache your login for weeks, and I can live with entering a PIN and my password now and then as a tradeoff for using something that's entirely hardware and software - independent. But of course, it'd be nice if it worked.
|
|
#
¿
Oct 6, 2022 20:53
|
|