LWN: Comments on "The endless browser wars"

The endless browser wars

immibis — Wed, 17 Feb 2021 12:52:09 +0000

I thought they meant to extract the API key from Chrome and use that.

Overreactions?

shane — Wed, 17 Feb 2021 10:16:26 +0000

I still have a i386 Netbook running Debian that I sometimes use for debugging my network. I wouldn't cry too much if Chromium went away as I mostly use Firefox but I do use occasionally Chromium to double-check if problems are browser-related or not.

The endless browser wars

ILMostro — Sun, 14 Feb 2021 18:35:49 +0000

Unfortunately, with more and more people becoming all too happy to forgo their privacy and freedoms in favour of technology that "just works", I'm not as optimistic about the adoption of "free alternatives" as I once was. I hope I'm wrong in this regard, for everyone's sake.

The endless browser wars

mathstuf — Tue, 09 Feb 2021 15:51:59 +0000

Oh, that's good news. I remember the list just having replies of "no new clients yet". I guess the GNOME/Epiphany team contacted them off-list then.

The endless browser wars

ceplm — Tue, 09 Feb 2021 01:44:31 +0000

That is not correct, actually. Epiphany supports Firefox Sync accounts.

The endless browser wars

emorrp1 — Mon, 08 Feb 2021 07:36:08 +0000

> The setup instructions make it very clear that us non-enterprise peasants, perhaps everyone outside of Mozilla, are not supposed to be even looking at this stuff, much less trying to use it.

I am a happy self-hoster of firefox-syncserver for the data, (but Firefox Account for just identity) and it's been pretty awesome to not worry about sending off my browser history and still be able to sling tabs between devices for reading on the go or at home.

The main issue is that despite some good efforts in 2017, it still hasn't been fully converted to python3 - so in the coming year I'll have to work out how best to keep running it following python2 removal from repositories. Hopefully someone else will do the hard work, then we can package it for debian (#900867) which would hopefully make running it yourself more accessible.

The endless browser wars

bblacksr — Sun, 07 Feb 2021 14:54:30 +0000

I just don't like Chromium or Chrome, unfortunately the only reason I use it is because my local paper stopped working in Firefox. Botheration.

The endless browser wars

iainn — Fri, 05 Feb 2021 22:53:08 +0000

Personally I'd recommend against a sync server that “does not persist data on restart”. That said, I recently read that Brave’s sync server is FOSS.

Overreactions?

flussence — Fri, 05 Feb 2021 20:00:13 +0000

Those patches are basically keeping KDE afloat in Gentoo right now.

It turns out making people spend 8-16 hours a day every 6 weeks compiling a half-gigabyte rendering engine that's often only used for frivolous HTML approximations of native widgets gets tiresome fast. Maybe abandoning QtWebkit was a bad idea.

The endless browser wars

linuxrocks123 — Thu, 04 Feb 2021 04:25:19 +0000

You can run your own Chromium Sync server if you want to.

https://github.com/ipernet/chromium-sync-server/

Mind you, I've never tried it. I run my own Weave 1.1 sync server and use Pale Moon.

Don't forget to call out websites which refuse to run on anything but Google Chrome (and Safari)

giraffedata — Sun, 31 Jan 2021 05:25:47 +0000

So you weren't willing to die for software freedom?

Don't forget to call out websites which refuse to run on anything but Google Chrome (and Safari)

jrw — Sat, 30 Jan 2021 02:16:24 +0000

Today I had to try 4 different browser/OS combinations before I found one which worked. I was trying to get in the virtual line for a Covid vaccine at my local supermarket chain (clinic.meijer.com/register). Their registration form workflow failed part-way through with Linux Mint/Firefox, Android/Firefox, and even Linux Mint/Google Chrome. It only worked with Chrome on Android.

I really try to avoid such websites, but in this case I felt it was worth it to compromise. However I also felt duty bound to encourage them to try harder to use a framework which supports Firefox and test against it, so I called Meijer's customer service desk to complain and also sent them an email.

Chromium is better sometimes

mss — Fri, 29 Jan 2021 17:55:18 +0000

This could be due to variety of reasons, from Windows users being used to getting a BSoD from time to time, Windows GPU drivers emulating certain known-problematic operations in software, to GPU vendors deprecating driver support for older chips quicker.

Chromium is better sometimes

sheepdestroyer — Fri, 29 Jan 2021 16:52:08 +0000

These are the same GPUs running hardware decoding by default on Windows.

API Key rotation

kpfleming — Thu, 28 Jan 2021 14:20:58 +0000

The article says that changing the API Keys would require 'force-upgrading' Chrome users... but that's already the case for the vast majority of users, since the browsers are updated quite often and upgrade themselves automatically.

I suspect Google could rotate the API keys every 3-4 months without much of a disturbance for 'normal' (Google channel) Chrome users.

Overreactions?

ibukanov — Thu, 28 Jan 2021 12:54:17 +0000

On utility of jumbo builds in Chromium - https://randomascii.wordpress.com/2020/03/30/big-project-...

Google announcing removal of jumbo support - https://groups.google.com/a/chromium.org/forum/#!topic/ch... . According to my colleague who knows people on Chromium team the primary driver for the removal was a Google manager who got very upset after trying to figure out why his patch broke a build. Eventually he figured out it was due to the jumbo feature that added some small and very easy to follow restrictions on C++ code, but if one does not know about them, it can take some time to figure things out.

For getting jumbo to work with Chromium again - https://twitter.com/pati_gallardo/status/1352587508375293952

Overreactions?

fulke — Thu, 28 Jan 2021 02:40:08 +0000

Interesting. Could you post any references?

Overreactions?

mathstuf — Wed, 27 Jan 2021 17:08:01 +0000

In local development, sure. But CI and (official) package builds are almost always done from-scratch.

The endless browser wars

Creideiki — Wed, 27 Jan 2021 09:57:58 +0000

If bookmark synchronisation really is such a killer app, I would have thought that setting up Floccus would be less philosophically and technically challenging than switching to an even more proprietary browser.

Overreactions?

caliloo — Wed, 27 Jan 2021 09:27:52 +0000

Same for me. I can’t stand the nag of google account password every time it starts. (Yes I don’t want it to remember it given the computer setup)

Overreactions?

smurf — Wed, 27 Jan 2021 08:56:29 +0000

That looks like a broken build system. Any builder worth its name should be designed to pick up where it stopped and simply continue.

The endless browser wars

jhhaller — Wed, 27 Jan 2021 02:04:16 +0000

On Mobile, Quora comments/answers don't work well in Firefox. Sometimes, when correcting, Firefox will delete much of the next word when typing a space. I've been trained to use Chrome to open Quora links. I'm sure their app works fine, I just don't want to use it.

Chromium is better sometimes

mss — Tue, 26 Jan 2021 23:39:45 +0000

It's not the VA-API itself that is unstable but the backing GPU and its drivers.
Often not only when decoding videos but also when doing just hardware-accelerated OpenGL rendering.

It looks like GPUs are designed primary with performance in mind, not long-term stability.

The endless browser wars

jmclnx — Tue, 26 Jan 2021 19:21:19 +0000

Funny, where I work on my RHEL Workstation, WEBEX is fine with chrome but has issues with Firefox. (note I prefer firefox).

The endless browser wars

mcatanzaro — Tue, 26 Jan 2021 18:25:34 +0000

To be clear: this is a separate issue. Since 2019, Google has required that API keys be kept secret. That is, they can no longer be included in open source projects (which presumably includes distro build systems). Everybody has just been ignoring the new rules since then. The Safe Browsing team clearly intends for their API to be used by open source projects, but it is just not possible to do anymore without violating the terms of service or somehow injecting secrets into the build process. Secrets are actually possible when building with GitLab (which is why the upstream build still has Safe Browsing enabled), but not possible for Linux distributions with fully public build systems and a goal of making builds reproducible.

The endless browser wars

mcatanzaro — Tue, 26 Jan 2021 18:20:41 +0000

Meanwhile, I've decided to disable Safe Browsing in Epiphany: https://gitlab.gnome.org/GNOME/epiphany/-/blob/74912f8ccb...

Firefox Sync is still going strong though. It's really amazingly nice of Mozilla to make this service available to non-Firefox apps with very few requirements (they boil down to "don't be malware" and "don't pretend to be Firefox").

The endless browser wars

NYKevin — Tue, 26 Jan 2021 16:17:32 +0000

Safe Browsing has been used by Firefox and Safari, as well as a number of smaller browsers, for years. Locking it down now would certainly be a surprise.

(Disclaimer: I work for Google. I don't work on Chrome, and don't know anything more about this than you probably do.)

The endless browser wars

mathstuf — Tue, 26 Jan 2021 14:46:45 +0000

> Perhaps someone is working on a patch to add support to the same 3rd party services Firefox users can use, and it hasn't been reported?

Mozilla is not handing out OAuth2 application keys to non-Mozilla projects for use with Mozilla's deployment. You can set up your own fxa-auth and fxa-sync servers though.

The endless browser wars

joey — Tue, 26 Jan 2021 14:23:32 +0000

For the distributions that care about software freedom, this is a form of technical debt raising its head. They have avoided confronting the problem of free software that depends on non-free network services, and so their users who want to avoid that are left with an infinity of config settings to try to accomplish it, and meanwhile they've helped hook many of their users on these non-free services rather than promoting alternatives from the beginning.

Overreactions?

mathstuf — Tue, 26 Jan 2021 14:17:56 +0000

True, but are there enough of you (I certainly dislike using Chromium and only use it where the site is completely busted in Firefox, so this stuff is of little relevance to me) to warrant the man-weeks it takes to handle the management of the package at the distro level? That's really the question.

The endless browser wars

al4711 — Tue, 26 Jan 2021 12:39:01 +0000

Finally Chromium starts to get more privacy friendly with removing the proprietary parts.

I hope this "Safe Browsing" tracking will also be remove as it is removed on this patch-set https://ungoogled-software.github.io/

The endless browser wars

timrichardson — Tue, 26 Jan 2021 11:43:42 +0000

I interpreted production use as the load of a distribution's Chromium users, all of whom use the same keys. If you use your own keys, the key has a user population of 1, and developer quotas may be ok for that. Maybe. I used my own keys with the saiarcot PPA (as described here https://launchpad.net/~saiarcot895/+archive/ubuntu/chromi...) for Ubuntu, which for a long time was the best way to use vaapi decoding on Ubuntu (the maintainer did not include any keys with his build). I didn't run into any quota problems. In the Google thread, the Google dev asking as spokesperson said a workflow based on individual user keys may be disabled one way or another in the future.

I hope that the actions taken by distribution maintainers to "disable APIs" still allow for BYO keys, although 95% of my browsing is in Firefox due to the Container Tabs feature which is miles ahead of Google profile functionality.

Chromium is better sometimes

juliank — Tue, 26 Jan 2021 11:21:56 +0000

It's not "worse", it's just that vaapi can be incredibly unstable and crash your system, and Google does not want to deal with the bug reports from that.

Overreactions?

ibukanov — Tue, 26 Jan 2021 10:56:00 +0000

At one point Opera contributed to Chromium a jumbo build option that reduced the build time from 8 hours to 1 and half roughly. Then a year ago Google removed those patches from Chromium tree as Google does not need them as they compile with their distributed server farm. But those patches are still maintained outside Chromium.

Chromium is better sometimes

sandsmark — Tue, 26 Jan 2021 09:08:53 +0000

> At the same time, Chromium in RPMFusion repositories carries patches to fix hardware video decoding, making it a better browser than plain Chrome.

Not anymore, it just needs to be enabled at build time. Arch Linux has enabled it without patches (AFAIK) for a fairly long time now: https://github.com/archlinux/svntogit-packages/blob/packa...

Overreactions?

LtWorf — Tue, 26 Jan 2021 08:54:51 +0000

For me it would certainly be a feature if it stopped to phone home to google.

The endless browser wars

rsidd — Tue, 26 Jan 2021 08:33:03 +0000

At least on linux, Cisco Webex works on firefox (via an addon) but not on chrome.

Chromium is better sometimes

zdzichu — Tue, 26 Jan 2021 06:49:47 +0000

I have an use-case where Chromium(-freeworld) is better than Chrome. I'm a casual gamer. So casual I play maybe 5-6 hours per month, so investing in dedicated gaming hardware is unreasonable. Instead I play on Stadia – Google's streaming gaming platform.

First of all, it works only with Chromium and derivatives, like MS Edge. It doesn't work with Firefox.

Second, the core requirement for gaming platforms is a realtime decoding if 1080p or 4k video streams. Chrome on Linux DOES NOT support accelerated video decoding. Software decoder in this browser introduces over 1 second of delay, making first person shooters unplayable.
At the same time, Chromium in RPMFusion repositories carries patches to fix hardware video decoding, making it a better browser than plain Chrome.

It's ironic that Google-provided browser is worse for Google platform than 3rd-party recompile.

(It is also ironic that games available on Stadia are being run on Linux, while there are not available on Linux as standalone)

The endless browser wars: Debian edition

calumapplepie — Tue, 26 Jan 2021 04:18:57 +0000

Debian was really ahead of the ball here. We've been disabling Google Sync in our chromium package for years. Whats more, we've already removed it from bullseye!

Jokes aside, Chromium packaging is a massive headache. It's been knocked out of bullseye because the maintainer didn't feel they could keep up alone, and the current factor blocking its re-entry is the fact that people are hesitant to commit to providing security support to such a headache of a package for the next 3 years. Chromium might be open-source, but it's really a Google product intended to be downloaded from Google and used under their watchful gaze. I can't say I'm shocked at this change.

The endless browser wars

pizza — Tue, 26 Jan 2021 00:54:58 +0000

MS Edge syncs (presumably) using some sort of Microsoft-specific protocol that is intimately tied to Microsoft's own account system.

Both Chrome and Edge are intended to help lock you into their respective maker's software-as-a-service ecosystems;

(To be blunt all of these "chromium" users put together probably don't even count as a rounding error vs Chrome or Edge, to say nothing of their combined install base..)