Firefox on Ubuntu (which I'm not using anymore for these and other reasons) and presumably on other OSs does exactly that: refuse to open new tabs/sites once every two weeks or so, telling me "one more thing we need to do" ie self-update. So you're loosing all your browsing context, though FF does a better job restoring it after restart than it used to, but still that's user-hostile and self-important as fuck. And there's also a "principal" reason why I can't stand this: that the Web is now 30 years old, past its peak, so if browsers still need to update bi-weekly for new features and experiments, this in combination with lack of browser diversity proves without any shade of doubt there's something very wrong with the incentives for browser development, the Google/Mozilla browser cartel, and the evolution of "web standards".
Nope, it only happens when something outside changed Firefox's files (in this case when Ubuntu swapped files because dpkg updated Firefox). This never happens* in Windows and macOS (it might nag, but you can definitely dismiss it). It seems that johnchristopher has a suggestion to disable auto-updates on a specific application in Debian and Ubuntu (haven't tested it though): https://news.ycombinator.com/item?id=33202052
* At least using their official installers. I'm specifically excluding using Chocolatey/brew/other loose-file update mechanism or your bonkers enterprise solution insists on using loose files and not rely on *.msi/*.app/*.pkg.
This is exactly right. And this happens because it needs to load libraries or exec subprocesses that aren't compatible between versions. Since the matching version of the file has been replaced with a newer one it doesn't really have a choice, it is unable to launch the new tab. The nice message is a better alternative than crashing.
This also doesn't occur on NixOS because the new version is in a different direcotry and the old version is kept until it is garbage collected.
> This also doesn't occur on NixOS because the new version is in a different direcotry and the old version is kept until it is garbage collected.
Also with flatpak; you are using the old version until you close it. Then it will be garbage collected. On the next launch, you will be running a new version from different root.
This is 100% something that can be fixed by the browser. Package updates don't change existing files but rather create new files and update the filesystem to point to them. All it takes to not need a restart is to keep file handles open for all needed resources and forking from such a process instead of exec()ing new ones. Not entirely trivial for applications as monstrously complex as browsers but hardly comparable to other challenges they need to solve. It's only because they can control the update process on the OS they care about (Windows) that they don't bother.
If you download the linux version of Firefox from mozilla.org you get a tarball that you can extract and run Firefox from without needing to do anything to install it. When I've run it this way it self updates the same way as it does on macos or windows.
> Nope, it only happens when something outside changed Firefox's files … This never happens* in Windows and macOS (it might nag, but you can definitely dismiss it).
Happens with multiple Firefox instances too, `firefox -P -no-remote` (I think in recent versions the -no-remote is redundant), even on Windows with official installers. In that case you don't even get the error message; new tabs just remain blank. There might be a delay between process A's update and symptoms appearing in process B; not sure.
> the Web is now 30 years old, past its peak, so if browsers still need to update bi-weekly for new features and experiments, this in combination with lack of browser diversity proves without any shade of doubt there's something very wrong with the incentives for browser development, the Google/Mozilla browser cartel, and the evolution of "web standards".
Or maybe new features are still coming out in the W3C specs and need to be implemented. Did you know CSS now has a parent selector, or that JS will be getting functional piping soon?
On functional piping, this has been in the pipeline with a few competing proposals for a while (I prefer the F#-like version myself). Will be surprised if/when it actually makes it in.
Yeah that is annoying, though personally i find browsers as the one category of software where i think this is fine, mainly because of security updates (it sucks that they tend to come with UI updates but at least on Firefox so far the main UI is fairly customizable and have made my own) but also because they are online software anyway.
Does a browser really have to be so complex that it warrants updates more often than the very sites that are being browsed? Contrast this with the design of idk MP3: a relatively simple and ultra-stable decoder app with a large variety of backend pipelines that can create MP3s. That's how the web pre-JS, pre-CSS was like. Or, with a perspective from information theory: downloading hundreds and hundreds of megabytes again and again (browsers), then consuming insane amounts of energy to access information that hasn't really changed all that much isn't very effective, is it?
>Does a browser really have to be so complex that it warrants updates more often than the very sites that are being browsed? Contrast this with the design of idk MP3: a relatively simple and ultra-stable decoder app with a large variety of backend pipelines that can create MP3s.
The problem is the recent trend that everything has to be a web application. So browsers aren't just to access information anymore, but literally to do everything else too. I personally don't agree with the web application trend, but this is the reason why a browser is so much more complex compared to an MP3 decoder: the decoder has to do a single thing, the browsers have to do more and more things.
Even your simple example, MP3 decoders, that do one thing, have had code execution exploits and other security issues over the years. WinAMP had CVEs for it.
All software will have bugs. I want my fixes fast and often to an environment where I run untrusted code from that many places.
Does it have to? I don’t know. But it is. It’s an operating system where 3rd parties execute random code on, and you hope it stays sandboxed. Those websites? Thanks to ads, they don’t update once in a while, but usually once every few seconds.
I just download it from Mozilla and install on my home dir. This solves both the issues of Debian not updating Firefox fast enough and of them updating it too fast.
This is infuriating to me as well. And the stupid Firefox restart dinosaur always comes up at the worst possible time. I would have switched to a new browser years ago, but Firefox seems to be the only one that supports vertical tabs.
Maybe I'm missing something or set some settings years ago and have forgotten but I have zero problems with restarting Firefox after an update from my package manager.
I regularly have 2-4 Firefox windows with dozens of tabs in each and a "You need to restart" button press takes like 4-5 seconds max to close all windows and reload them all with all of my tabs as they were. The most I have to do is stick each instance on the right workspace.
Granted each tab will reload when opening it but if I'm updating my OS packages I'm probably not exactly "in the zone".
I am using opera with “tree tabs” extension, but afair there is a similar (or the same?) extension for chrome. Although they do not hide the horizontal tab bar.
Edge is not free. I wouldn't seriously consider using a non-free application for something as essential as everyday web browsing.
You can't really have usable vertical tabs in Chromium via plugins either, unless you're content with wasting a lot of horizontal space for an ugly sidebar and vertical space for uselessly duplicated tab bar.
Firefox is the only actual choice I'm aware about.
Using Firefox definitely supports the existence of a free browser. Loss of market share is the #1 threat to the continued existence of a free browser. Beyond the obvious (if a tree falls in a forest, crushing the last copy of the code for a browser that has zero users, then was it a browser at all?):
lower market share =>
nobody testing against the free browser or fixing site breakage =>
quirks (bugs, underdefined specifications, nonstandard features) of other browsers becoming required for a functional Web =>
free browser is no longer a browser of the actual Web.
Being controlled by corporate interests is completely orthogonal to being free. A lot of Free Software is being controlled by corporate interests and there's nothing wrong with it.
Given that the issue of Firefox being forced to restart primarily happens on Linux, I doubt Edge is an option for them. Though I have to concur that Edge has one of the most stable and smooth vertical tab implementations around, most of the plugin-based ones are more fully featured but much less reliable.
I use it, and it's decent. And more in the vein of "it's not google" though I do slightly prefer the chrome dev tools to the modifications that Edge has made. I don't like a lot of the "helpers" for shopping though. And definitely don't like the article wall with ads that are really hard to block/script out.
The first time it happened, I would get rid of whatever software was involved in causing it and never use it again, except for testing purposes.
I completely agree with you about everything else too. The Web is mature enough that I can use a well-tested website with a 20-year-old browser.
There is no technical reason to not have a minimum-viable web browser with a smaller attack surface that doesn't need upgrades for months or even years.
And, in fact, such browsers exist, and I can browse most of the Web that I need with them. I just have to ignore the shitty mainstream, which I am more than happy to do.
As someone else pointed out above. This is not due to Firefox per se but due to the OS changing dependent packages. I understand if you install Firefox directly from their website rather than through the system package installed you won’t have that behavior. It does not happen on other OSs.
Firefox on Mac. I have this problem as well, even with "autoupdate" set to true. Also, if I am browsing in the middle of an auto update, it refuses to open new tabs.
And each update logs me out of my password manager
Worse, now this is happening to Thunderbird. Wasn't an issue until about a month ago. Now I need to re-install Thunderbird every few weeks because they've pushed an update.
When from my point of view, as a user, I haven't seen a new feature which I truly wanted in over 5 years, maybe more like 10.
I use FF on Mac OS and have never seen it refuse to operate due to an update. It will sometimes popup a dialog letting me know there is an update but I can dismiss it. I tend to check for updates every couple of weeks, anyway. The update process is mostly painless though it does restart the browser. That is probably a good thing anyway. Most updates I get when I do update the browser for other reasons and it installs and update as it is restarting.
That sounds truly horrible. I've only recently returned to Mac OS as primary OS (one reason being annoyed by Ubuntu and FF) but why aren't you using Safari and Mail.app then if I may ask?
That this complaint is pointing at Firefox when it's not Firefox at all, is interesting. How many things do we associate with one thing when it's actually something else entirely.
What does then happen? Do I get some nudge to update?
I for one don't really mind the Autoupdate. I can't stand it that it forces me to restart while I am in some workflow. I'd be fine with "you should update, click here, when ready" which I can click five minutes later when done with the task.
This reminds me of iOS apps that as soon as you open them require an update to the latest version and won’t let you proceed any further (in the app) until you click OK/do the update. I always make it a point to go leave a one star review (or update my existing review to one star), and point this out as the reason
