LWN: Comments on "A 2018 retrospective" https://lwn.net/Articles/775050/ This is a special feed containing comments posted to the individual LWN article titled "A 2018 retrospective". en-us Wed, 15 Oct 2025 18:52:14 +0000 Wed, 15 Oct 2025 18:52:14 +0000 https://www.rssboard.org/rss-specification lwn@lwn.net A 2018 retrospective https://lwn.net/Articles/776540/ https://lwn.net/Articles/776540/ nix <blockquote> Mobile Firefox is much slower than mobile Chrome </blockquote> And then you install an adblocker, and mobile Firefox is suddenly blindingly fast because you're not blocking waiting for massive piles of crap on every webpage. (I installed it because I <i>had</i> to: on a 10GiB/month satellite connection you do not want to be downloading 10MiB of useless advertising rubbish and horrible uninteresting autoplaying videos on every other page you visit.) Fri, 11 Jan 2019 16:40:59 +0000 Firefox https://lwn.net/Articles/775900/ https://lwn.net/Articles/775900/ sdalley <div class="FormattedComment"> If it's after cold startup, maybe it's something in the video/graphics stack which isn't initialized properly, and firefox is the first thing to trip over it.<br> </div> Thu, 03 Jan 2019 12:25:23 +0000 Firefox https://lwn.net/Articles/775807/ https://lwn.net/Articles/775807/ pizza <div class="FormattedComment"> You want a stable base platform, UI, and plugin API? Try Mozilla Seamonkey. Barely changed for the better part of two decades.<br> <p> And a market share that's little more than a rounding error of a rounding error. So much to stability mattering.<br> <p> <p> </div> Mon, 31 Dec 2018 23:02:20 +0000 Firefox https://lwn.net/Articles/775799/ https://lwn.net/Articles/775799/ Cyberax <div class="FormattedComment"> <font class="QuotedText">&gt; Suggesting third parties can write extensions to handle pretty common user goals is not terribly compelling now that firefox has broken extension compatability so many times.</font><br> Uhm. The new extension API is by design stable and doesn't require plugins to access the inner workings of FireFox.<br> </div> Mon, 31 Dec 2018 20:47:38 +0000 Firefox https://lwn.net/Articles/775798/ https://lwn.net/Articles/775798/ k8to <div class="FormattedComment"> The method used to manage text areas involves installing a custom webserver to act as an addon-specific enabler, to execute commands on http request. It's more secure for people to run cobbled together http servers? Surely it would be vastly more secure for the browser to simply have a editor configuration setting.<br> <p> Suggesting third parties can write extensions to handle pretty common user goals is not terribly compelling now that firefox has broken extension compatability so many times.<br> <p> You're drawing a very narrow timeline to support your claims. For example, maybe this year maybe the firefox UI hasn't been redesigned. I still don't recognize the UI Firefox gives me today. The experience is wildly inconsistent across platforms (worse than in the past), with highly strange decisions like using a hamburger menu when there's a full menu bar across the screen. If firefox had some consistency with the base platform or itself, and was stable in UI operation for a few more years, then I'd believe there's been a change in priorities.<br> </div> Mon, 31 Dec 2018 20:38:53 +0000 Firefox https://lwn.net/Articles/775742/ https://lwn.net/Articles/775742/ flussence <div class="FormattedComment"> It sounds that way, but given the rest of the system has no problems compiling itself on a regular basis I doubt it…<br> </div> Mon, 31 Dec 2018 05:12:00 +0000 Firefox https://lwn.net/Articles/775738/ https://lwn.net/Articles/775738/ apoelstra <div class="FormattedComment"> But now Ctrl+V is right next to Ctrl+W, which is even worse! :)<br> </div> Sun, 30 Dec 2018 17:41:31 +0000 Firefox https://lwn.net/Articles/775713/ https://lwn.net/Articles/775713/ halla <div class="FormattedComment"> I'd run a memory checker if I were you, because this is very obviously not normal: something on your system is broken. When I has similar problems, it did turn out to be a rotten memory stick.<br> </div> Sat, 29 Dec 2018 11:41:20 +0000 Firefox https://lwn.net/Articles/775703/ https://lwn.net/Articles/775703/ flussence <div class="FormattedComment"> Regarding #2, Firefox does a few dirty mmap/IO tricks with its resource files to reduce its runtime RAM usage. E17 is another piece of software doing those things, with similar problems.<br> <p> For me the crashes usually come at cold startup. It's about a 50/50 chance whether I'll get a browser window or a crash handler window each day.<br> </div> Sat, 29 Dec 2018 04:54:40 +0000 A 2018 retrospective https://lwn.net/Articles/775689/ https://lwn.net/Articles/775689/ buck <div class="FormattedComment"> The Linus hiatus and its offshoots could plausibly be considered as instances of the open source community doing some introspection, if only in regards to the standards it sets for its culture and not necessarily for the standards that its products set for the world, though one can imagine that the former will rub off on the latter (though, as you like to say, "how, it remains to be seen"). The community/contributior guidelines/standards/codes of conduct that open source projects adopt, to the extent that they signal a cultural shift, is significant in and of itself, though, insofar as the open source development community is a prominent Internet "commons" that can be a beacon for online communities impregnable to subversion by the malefactors riving the "social network" fabric with controversy and scandal, by contrast.<br> <p> The cloud provider compromise prediction could be validated if Bloomberg's Super Micro story is somehow vindicated, though everybody but Bloomberg is probably rooting against you on that one.<br> </div> Fri, 28 Dec 2018 22:09:10 +0000 Firefox https://lwn.net/Articles/775676/ https://lwn.net/Articles/775676/ jezuch <div class="FormattedComment"> Chromium used to have a ctrl+shift+q shortcut to close, but apparently this was still too easy to hit so recent releases prompt you to go through the menus instead. Well, whatever, it's not mission critical after all...<br> <p> (It reminds me of the time when I played Quake using the keyboard and tried to simultaneously shoot (ctrl), strafe (alt) and lower my crosshair (del). This was interpreted by the program as the "quit" command. I decided then I had enough for the day.)<br> </div> Fri, 28 Dec 2018 18:48:32 +0000 Firefox https://lwn.net/Articles/775673/ https://lwn.net/Articles/775673/ Lekensteyn I used to have the Disable Ctrl-Q Shortcut extension installed, but this stopped working since Quantum. Since then I always have a tab open with this URL: <pre>data:text/html,&lt;body onbeforeunload=return!0&gt;&lt;title&gt;Disable Ctrl-Q</pre> Whenever I accidentally press Ctrl + Q (which has happened a non-zero number of times), it will prompt before closing the page (and by extension, before closing the browser). Fri, 28 Dec 2018 16:45:21 +0000 Firefox https://lwn.net/Articles/775672/ https://lwn.net/Articles/775672/ mathstuf <div class="FormattedComment"> Ah, I guess so. I probably skipped over that solution since "window" isn't a verb :) .<br> </div> Fri, 28 Dec 2018 16:27:57 +0000 Firefox https://lwn.net/Articles/775664/ https://lwn.net/Articles/775664/ halla <div class="FormattedComment"> I am pretty sure the W in Ctrl-W stands for "Window", and in an MDI application it will close a window.<br> </div> Fri, 28 Dec 2018 16:13:23 +0000 Firefox https://lwn.net/Articles/775660/ https://lwn.net/Articles/775660/ mathstuf <div class="FormattedComment"> C-q staying the same makes sense, but C-w sounds like one of those bindings you want to be scan code based (like A-Tab and A-~) since it isn't a mnemonic.<br> </div> Fri, 28 Dec 2018 15:10:44 +0000 Firefox https://lwn.net/Articles/775647/ https://lwn.net/Articles/775647/ nevets <div class="FormattedComment"> This is the first I heard about the Ctrl-W vs Crtl-Q issue. But then again, I use a Dvorak layout where Q and W are far apart ;-)<br> </div> Fri, 28 Dec 2018 02:22:07 +0000 Firefox https://lwn.net/Articles/775569/ https://lwn.net/Articles/775569/ garrison Oh interesting. It seems the warning/confirmation I desire <a href="https://github.com/mozilla/gecko-dev/commit/db2b40265a0237a2022313bf42d80f4e6357857b">has hit master in the past month</a>. Together with the fix of crashing after upgrade (as mentioned elsewhere in this thread, it is fixed in Firefox 64), I may actually be switching to Firefox in 2019. My best guess is that this will land in Firefox 66? Marking my calendar for March 19 to try Firefox again. Tue, 25 Dec 2018 04:33:43 +0000 Proton sea change https://lwn.net/Articles/775566/ https://lwn.net/Articles/775566/ flussence <div class="FormattedComment"> MS is having a hard time even giving Windows away gratis in the newly post-Ballmer era, so they're mostly counting on being a service and SaaS provider (Office 365 etc) for revenue. They also make money off the games division (or at least, I believe it's not $5bn in the red any more…)<br> </div> Mon, 24 Dec 2018 23:45:45 +0000 Proton sea change https://lwn.net/Articles/775556/ https://lwn.net/Articles/775556/ raven667 <div class="FormattedComment"> What would be funny would be for MS to work on WINE as a way to preserve Win32 software as they probably can't relicense every bit of their legacy code to open source and WINE license is free and clear. It would be a weird world where WSL is a backup implementation of Linux and WINE of Windows. Is MS today a bigger data center hosting company than software?<br> </div> Mon, 24 Dec 2018 18:55:18 +0000 Firefox https://lwn.net/Articles/775554/ https://lwn.net/Articles/775554/ garrison <div class="FormattedComment"> Another issue is that I use awesomewm, and I put a fair amount of cumulative effort into making sure each browser window is associated with the appropriate awesomewm tag(s). Even if all my open Firefox windows and tabs are restored after I accidentally press Ctrl+Q, I still need to go through and manually assign them all to the intended awesomewm tags again. So claiming there is "no risk of data-loss, and therefore no need to warn" is not consistent with my experience.<br> </div> Mon, 24 Dec 2018 18:20:02 +0000 Proton sea change https://lwn.net/Articles/775451/ https://lwn.net/Articles/775451/ jeffcook <div class="FormattedComment"> Always expected something like Proton eventually, but the timing is an important indicator of Valve's thought process: it shows that Valve is at least _somewhat_ confident that Microsoft will allow large-scale WINE rollouts from well-known companies without unleashing an unholy lawsuit apocalypse.<br> <p> CodeWeavers, Cedega, et al have always been too small to bother launching a big lawsuit against -- Streisand Effect would've done more damage than those groups posed -- but Valve pushing this is serious business.<br> <p> Whatever the outcome, it marks a crucial milestone, and IMO shows that Microsoft may have actually internalized a post-Windows future for itself.<br> </div> Sat, 22 Dec 2018 02:46:54 +0000 Firefox https://lwn.net/Articles/775446/ https://lwn.net/Articles/775446/ roc <div class="FormattedComment"> <font class="QuotedText">&gt; Any infrastructure that allows users to avoid data loss in textareas now that addons are dead would be a good start.</font><br> <p> Addons aren't dead and insisting on saying that they are just makes you look foolish. For example there are many WebExtension addons to manage textarea contents, e.g.:<br> <a href="https://addons.mozilla.org/en-US/firefox/addon/formsave/">https://addons.mozilla.org/en-US/firefox/addon/formsave/</a><br> Took me about five seconds to Google that.<br> <p> <font class="QuotedText">&gt; Functionality to make sites more accessible by expanding the box where the text is at the expense of the chrome on all sides of the screen would be not TOO hard to deploy.</font><br> <p> Would be easy in a WebExtensions addon if one doesn't already exist.<br> <p> <font class="QuotedText">&gt; Safety things like "this domain seems to be constantly changing certs" would be readily implemented but aren't.</font><br> <p> Maybe your favourite "safety thing" isn't implemented but lots of security work keeps getting done. Better sandboxing for example.<br> <p> <font class="QuotedText">&gt; Basically the entire effort of enormous amounts of engineering time is spent only on adding developer focused features.</font><br> <p> That's not true at all. For example lots of work is going into machinery and UI for blocking trackers and other stuff.<br> <p> It's true that an enormous amount of work is going into Web-developer-facing features (both implementing and standards work), but that's table stakes for Gecko to be a viable independent browser engine, without which the open Web is doomed.<br> <p> An enormous amount of work goes into performance too, which is the most important user feature.<br> <p> There's actually not much work going into UI reorganization currently AFAICT. Funnily enough Chrome just revamped their UI to look more like Firefox.<br> </div> Fri, 21 Dec 2018 22:55:33 +0000 Firefox https://lwn.net/Articles/775443/ https://lwn.net/Articles/775443/ k8to <div class="FormattedComment"> I'm very familiar with all the many trade-offs that have been made. Sound technical reasons are a common way that engineers screw over users in open source and in corporate development.<br> <p> However, there's many simple wins that could exist. Any infrastructure that allows users to avoid data loss in textareas now that addons are dead would be a good start. Functionality to make sites more accessible by expanding the box where the text is at the expense of the chrome on all sides of the screen would be not TOO hard to deploy. Safety things like "this domain seems to be constantly changing certs" would be readily implemented but aren't. <br> <p> Basically the entire effort of enormous amounts of engineering time is spent only on adding developer focused features. That's a strategy for sure, but it's not convincing to me.<br> <p> Meanwhile a very large amount of engineering time is spent on reorganizing the UI over and over, defeating user learning. There's a good rule of thumb on UI redesigns. If you're making the UI 2x as good or better (how much effort is required to do the common tasks), go ahead and redesign, otherwise it's not worth it. Firefox isn't anywhere near this good maxim. Chrome is doing a better job here, which is a pretty low bar.<br> <p> Yes, a browser monoculture is bad, but multiple browsers would be better if any of them were showing sanity in the face of these types of things.<br> </div> Fri, 21 Dec 2018 21:49:36 +0000 Firefox https://lwn.net/Articles/775436/ https://lwn.net/Articles/775436/ roc <div class="FormattedComment"> The addons model was changed for sound technical reasons.<br> <p> I can't even imagine the conspiracy-theory explanation that makes dropping XUL addons some kind of evil plot.<br> </div> Fri, 21 Dec 2018 19:19:04 +0000 Firefox https://lwn.net/Articles/775423/ https://lwn.net/Articles/775423/ garrison From the link: <blockquote> <blockquote>Is there a follow-up plan here to add a warning when closing Firefox even with session restore on? </blockquote> <p>In short: no plan for that. </p> <p> Reason: Firefox saves everything, so there is no risk of data-loss, and therefor no need to warn.</p> </blockquote> <p>Seems short-sighted to me. What if I have an incognito window open? It's certainly not coming back after the restore.</p> Fri, 21 Dec 2018 15:29:14 +0000 Firefox https://lwn.net/Articles/775390/ https://lwn.net/Articles/775390/ excors <div class="FormattedComment"> That's a pretty big "if" :-)<br> </div> Fri, 21 Dec 2018 13:17:54 +0000 Firefox https://lwn.net/Articles/775389/ https://lwn.net/Articles/775389/ excors <div class="FormattedComment"> showQuitWarning was removed (as of Firefox 63, I think): <a href="https://bugzilla.mozilla.org/show_bug.cgi?id=1438499#c14">https://bugzilla.mozilla.org/show_bug.cgi?id=1438499#c14</a> looks like a relevant comment. And the remaining options intentionally don't warn on quitting if you have session restore enabled: <a href="https://bugzilla.mozilla.org/show_bug.cgi?id=1438499#c34">https://bugzilla.mozilla.org/show_bug.cgi?id=1438499#c34</a><br> </div> Fri, 21 Dec 2018 13:16:02 +0000 Firefox https://lwn.net/Articles/775388/ https://lwn.net/Articles/775388/ RX14 <div class="FormattedComment"> Note, you need to disable "restore previous session" in the settings to make the ctrl-q fix work. See <a href="https://superuser.com/a/1352295/285546">https://superuser.com/a/1352295/285546</a><br> </div> Fri, 21 Dec 2018 13:00:03 +0000 Firefox https://lwn.net/Articles/775387/ https://lwn.net/Articles/775387/ RX14 <div class="FormattedComment"> Set browser.showQuitWarning=true and browser.warnOnQuit=true in about:config to solve ctrl-q<br> <p> The "firefox crashes after an apt-get upgrade" issue is also fixed in firefox 64.<br> <p> I suggest you try firefox again!<br> </div> Fri, 21 Dec 2018 12:58:43 +0000 Firefox https://lwn.net/Articles/775382/ https://lwn.net/Articles/775382/ pizza <div class="FormattedComment"> <font class="QuotedText">&gt; None of the browsers are really for user goals at all anymore. That happened a long time ago.</font><br> <p> If "user goals" are what you truly care about, then even at its worst, Firefox remains vastly superior to its alternatives.<br> <p> (Granted, there are some FF derivatives out there that may be "better" but if FF itself goes away, so will they..)<br> <p> </div> Fri, 21 Dec 2018 11:36:42 +0000 A 2018 retrospective https://lwn.net/Articles/775377/ https://lwn.net/Articles/775377/ marcH <div class="FormattedComment"> <font class="QuotedText">&gt; Windows is more likely to end up as a new Linux distro with its own very different behavior, rather than actually being related to Ubuntu. If you separate behavior into policy and mechanism</font><br> <p> I don't have a clue what that means. On the other hand I use Microsoft's subsystem for Linux all the time and I can attest that: 1) it is outstanding work 2) it doesn't really care which Linux distro you use - no more than Linux containers do.<br> <p> </div> Fri, 21 Dec 2018 09:08:42 +0000 A 2018 retrospective https://lwn.net/Articles/775376/ https://lwn.net/Articles/775376/ marcH <div class="FormattedComment"> I think it was before 2018 but I still cannot believe how fast and easy it was to apply this HOWTO to my router and workaround the bufferbloat in my cable modem:<br> <p> <a href="https://openwrt.org/docs/guide-user/network/traffic-shaping/sqm">https://openwrt.org/docs/guide-user/network/traffic-shapi...</a><br> <p> Thanks for all the hard work, well done.<br> <p> <p> </div> Fri, 21 Dec 2018 09:02:24 +0000 Firefox https://lwn.net/Articles/775375/ https://lwn.net/Articles/775375/ marcH <div class="FormattedComment"> <font class="QuotedText">&gt; don't write long comments directly into web pages (write in a separate text editor then copy-and-paste),</font><br> <p> They say emotions good or bad are what trains memory. A lot of pain is for sure what taught me to do this systematically.<br> <p> As this habit "accidentally" extended to cloud documents, an interesting side-effect is to stop the distraction of starting to pretty the form before the content is mostly ready.<br> <p> </div> Fri, 21 Dec 2018 08:58:51 +0000 Firefox https://lwn.net/Articles/775372/ https://lwn.net/Articles/775372/ k8to <div class="FormattedComment"> And this scenario where firefox killed all their own addons is why I am not much interested in Firefox anymore.<br> <p> None of the browsers are really for user goals at all anymore. That happened a long time ago.<br> </div> Fri, 21 Dec 2018 07:37:14 +0000 A 2018 retrospective https://lwn.net/Articles/775365/ https://lwn.net/Articles/775365/ dskoll <div class="FormattedComment"> I've always used Firefox on my desktop boxes. I switched my Android devices to Firefox also. Mobile Firefox is much slower than mobile Chrome, even though desktop Firefox seems about as fast as desktop Chrome.<br> <p> Nevertheless, I will stick with Firefox because it's too important not to yield complete control to Google.<br> </div> Thu, 20 Dec 2018 23:29:20 +0000 Firefox https://lwn.net/Articles/775362/ https://lwn.net/Articles/775362/ mcatanzaro <div class="FormattedComment"> So what is the complaint about then... hmm... OK, I see Chrome doesn't respect Ctrl+Q at all. That's pretty wild; you'd think that'd be the one standard shortcut all desktop apps would understand....<br> </div> Thu, 20 Dec 2018 21:34:53 +0000 Firefox https://lwn.net/Articles/775359/ https://lwn.net/Articles/775359/ roc <div class="FormattedComment"> If the Web page doesn't suck then restoring it after close in Firefox should restore any form control state as well.<br> </div> Thu, 20 Dec 2018 21:19:02 +0000 Firefox https://lwn.net/Articles/775358/ https://lwn.net/Articles/775358/ roc <div class="FormattedComment"> Firefox remembers tabs on quit unless you configure that off.<br> </div> Thu, 20 Dec 2018 20:43:23 +0000 Firefox https://lwn.net/Articles/775357/ https://lwn.net/Articles/775357/ roc <div class="FormattedComment"> I wonder if #2 is a distro issue. Try running a Mozilla build of Firefox and see if it has the same problem.<br> </div> Thu, 20 Dec 2018 20:42:54 +0000 Firefox https://lwn.net/Articles/775354/ https://lwn.net/Articles/775354/ mcatanzaro <div class="FormattedComment"> Epiphany: still around, and at least capable of remembering your tabs when you hit Ctrl+Q....<br> </div> Thu, 20 Dec 2018 20:00:39 +0000