Firefox 52.0

Posted Mar 8, 2017 12:59 UTC (Wed) by iive (guest, #59638)
Parent article: Firefox 52.0

PulseAudio is now the only audio output enabled by default.

Just upgraded to Firefox 52 and all HTML5 videos got silent.

I do keep my system PA clean. For selected programs I do use a-pulse project.
Unfortunately the last release of a-pulse (0.1.7) doesn't work with Firefox 52, you'd need the git version (no volume control yet).

If I understand correctly the following thread, ALSA support could still be enabled at compile time:
https://bugzilla.mozilla.org/show_bug.cgi?id=1247056
I'll ask my distro to enable it.

Just to point the obvious. PulseAudio does depend on ALSA to access the sound card.

Firefox 52.0

Posted Mar 9, 2017 18:26 UTC (Thu) by tuna (guest, #44480) [Link] (22 responses)

Thank you for your amusing anecdote.

Firefox 52.0

Posted Mar 9, 2017 19:37 UTC (Thu) by sbergman27 (guest, #10767) [Link] (21 responses)

Not amusing. All these years later, Pulse is *still* causing unnecessary strife for some. And all because some of the cheapest audio cards of the 1990s couldn't do hardware mixing of streams. I disable Pulse, as well.

Firefox 52.0

Posted Mar 9, 2017 19:59 UTC (Thu) by rahulsundaram (subscriber, #21946) [Link]

"And all because some of the cheapest audio cards of the 1990s couldn't do hardware mixing of streams"

That's not why it is default in so many distros.

Firefox 52.0

Posted Mar 10, 2017 14:48 UTC (Fri) by pizza (subscriber, #46) [Link] (19 responses)

s/some of the cheapest cards of the 90s/all but a handful of audio devices ever sold, including today/

Firefox 52.0

Posted Mar 10, 2017 20:34 UTC (Fri) by sbergman27 (guest, #10767) [Link] (18 responses)

Quite untrue. Even in the 90s I never had a card/driver combo that couldn't mix streams without a sound server. And I've supported a lot of sound cards over the decades.

Firefox 52.0

Posted Mar 10, 2017 22:11 UTC (Fri) by cortana (subscriber, #24596) [Link] (17 responses)

Sounds like you only bought the most expensive cards. I owned many SoundBlaster 64s, SoundBlaster 128s and various onboard sound cards and none of them supported hardware mixing.

Firefox 52.0

Posted Mar 11, 2017 13:22 UTC (Sat) by nix (subscriber, #2304) [Link] (16 responses)

Quite. And since then, what *everyone* has, because it's on essentially every non-server motherboard out there and even a few server mobos, is Intel HDA. And most implementations of that don't support hardware mixing.

The last mixing-capable card I had was back in the late 90s. I don't miss it: all its features are implementable in software (PA, Timidity, etc) and the HDA sounds every bit as good as the more expensive mixing-capable card did.

Firefox 52.0

Posted Mar 11, 2017 19:38 UTC (Sat) by ajmacleod (guest, #1729) [Link] (15 responses)

I also suffered from horrible sound cards for a long time... the good news is that ALSA can do software mixing too (granted, only for the past decade or so IIRC, but...)

Firefox 52.0

Posted Mar 11, 2017 22:54 UTC (Sat) by zlynx (guest, #2285) [Link] (14 responses)

ALSA dmix?

It has ALL the problems of sound servers, does a crappy job of mixing, and has worse latency than PulseAudio, at least when PA is asked to provide low latency instead of power efficiency.

Firefox 52.0

Posted Mar 13, 2017 14:26 UTC (Mon) by ajmacleod (guest, #1729) [Link] (13 responses)

Even in theory it seems unlikely that adding another layer of bloat is going to be more efficient, but more to the point my own experience is that ALSA's own mixing is far more reliable and efficient than with PA added to the equation; with PA my sound was often choppy and late - when it worked.

Firefox 52.0

Posted Mar 13, 2017 15:15 UTC (Mon) by pizza (subscriber, #46) [Link] (10 responses)

...And I can add my own anectdotes about problems I've experienced with ALSA's dmix plugin and how PA worked better, even in its early days.

And incidentally, ALSA's dmix plugin, by your definition, is "another layer of bloat"

Firefox 52.0

Posted Mar 13, 2017 15:44 UTC (Mon) by anselm (subscriber, #2796) [Link]

It doesn't matter how well PulseAudio works today. It does solve real-world problems that people have, and in my experience it is fine. (For example, my mobile phone uses it and has no problems moving audio streams between the device and a Bluetooth headset even in mid-call.) There was, however, a time when PulseAudio (on Ubuntu) worked really badly, mostly due to bugs in the underlying ALSA drivers, and even though that wasn't really PulseAudio's fault in the first place it gets trotted out again and again whenever someone mentions PulseAudio or systemd or Lennart Poettering or anything that people think is new and therefore bad. “Remember PulseAudio!” is the generic anti-innovation argument.

It is a well-known fact that software never improves from the point in time when it was first looked at. Therefore to some people something like System-V init is perfect because we've already had 25 years to shake out the bugs, but something like PulseAudio is an unmitigated disaster that is not worth one's time even 10 years later, no matter how much work other people have put into it in the meantime.

Firefox 52.0

Posted Mar 14, 2017 23:42 UTC (Tue) by ajmacleod (guest, #1729) [Link] (8 responses)

DMix is part of ALSA, not an entire kitchen-sink-and-the-toilet-too additional sound server. The difference is obvious; is the fact that audio on my system works perfectly well without PA and that I therefore don't want to be obliged to use PA it really so offensive?

Firefox 52.0

Posted Mar 15, 2017 3:27 UTC (Wed) by pizza (subscriber, #46) [Link]

> is the fact that audio on my system works perfectly well without PA and that I therefore don't want to be obliged to use PA it really so offensive?

No, what is offensive is the expectation that other folks should write, support, or otherwise maintain software for you.

Firefox 52.0

Posted Mar 15, 2017 8:53 UTC (Wed) by anselm (subscriber, #2796) [Link] (6 responses)

is the fact that audio on my system works perfectly well without PA and that I therefore don't want to be obliged to use PA it really so offensive?

Being forced to use PulseAudio may be offensive to you but being forced to not use PulseAudio may be offensive to Firefox developers. Guess who has more clout as far as the development of Firefox is concerned?

Firefox 52.0

Posted Mar 15, 2017 15:19 UTC (Wed) by ajmacleod (guest, #1729) [Link] (5 responses)

Oh, I'm happy to admit that they have all the clout. What they won't have at this rate is users and hence significance. FF marketshare and "brand awareness" is already rapidly disappearing; the more they disappoint or annoy their long term users and promoters in the community, the faster that will happen.

Firefox 52.0

Posted Mar 15, 2017 16:03 UTC (Wed) by pizza (subscriber, #46) [Link] (3 responses)

While FF's marketshare and "brand awareness" are rapidly disappearing, it has nothing to do with users like you or I, running a niche OS comprising 2% of the theoretical userbase.

FF is losing on Windows to IE/Edge, on Android to Chrome, on Apple to Safari/IOS, because the default is MoreThanGoodEnough(tm) -- and it's main differentiating feature these days, (freedom/principles/whatever) doesn't actually matter to the 98%, at least not when compared to convenience, proprietary plugins, and having to deal with breakages caused by folks not testing on anything other than Webkit-derived browsers.

This is the old MS/Netscape battle all over again..

No amount of promotion by the likes of you or I will make one iota of difference in the face of that tide.

Firefox 52.0

Posted Mar 15, 2017 22:17 UTC (Wed) by ajmacleod (guest, #1729) [Link] (2 responses)

The thing is that the only reason FireFox ever really gained any marketshare in the first place was because of people like us who got other, less technically minded, people using it. I am responsible for the desktop that a lot of users see, and despite most of my users being on Windows they aren't using IE or Edge because they won't be able to find them easily; they have the browser of my choice set as their default (well, in between MS changing it every second "update", but..)

I would disagree then that FF's brand awareness or user base has nothing to do with me - it does, because without me an awful lot of people (clearly not millions, but at least several hundred) would probably never have used FF in the first place.

I know this particular change does not affect Windows users (at least, not on anything post XP) but it's the final straw for me in the way Mozilla has managed/handled/"interacted" with their user community which means I'll be switching most of my Windows based users from here on too.

Simply locking down the bug to make sure the real scale of the frustration at this decision is not seen is a prime example of their approach, and I don't like it; I'm not going to issue death threats, I don't personally have the competence to rewrite FF sound handling on Linux so I'll just go elsewhere and leave them to their echo chamber.

Firefox 52.0

Posted Mar 16, 2017 15:18 UTC (Thu) by pizza (subscriber, #46) [Link] (1 responses)

> The thing is that the only reason FireFox ever really gained any marketshare in the first place was because of people like us who got other, less technically minded, people using it

....And the plain fact that it was objectively _Much Much Much Much_ better than IE, which by that point had stagnated for many years. That was the key to Firefox's massive uptake.

Today, Firefox is more or less on par with everything else, ahead in some respects, behind in others. However, Mozilla still lacks the platform bundling advantage (and free "download Firefox" blurbs on half the web...) which puts it at a considerable disavantage in the long run.

And yes, how Firefox handles sound on Linux doesn't matter one bit.

Incidently, what are you going to do instead? Switch to Chrome, which is objectively much worse on the social/freedom/ignoring users scale?

Firefox 52.0

Posted Mar 16, 2017 15:42 UTC (Thu) by ajmacleod (guest, #1729) [Link]

Even when FF was genuinely miles better than the alternative, most users didn't really know how or even particularly care - in those days in fact there were still plenty of horrible websites out there which didn't work well or sometimes at all with anything other than IE, and a non-technical user would just assume FF was to blame.

Personally, I'm looking at PaleMoon again. I'm definitely not 100% happy about switching, in fact I'd much rather not; but at this stage I think it's the least bad option.

It's a pity Opera hadn't done something more than reskin Chrome, that might have been worth considering - Vivaldi was too awful for words on the few occasions I've tried it.

For other people's Windows desktops (and possibly those on Linux) it'll be Chrome from here on in - it works about as well as Firefox on Windows, privacy & monoculture issues excepted. If Mozilla aren't interested keeping their users happy I don't see why I should assist them in their battle for relevance.

Firefox 52.0

Posted Mar 15, 2017 16:21 UTC (Wed) by anselm (subscriber, #2796) [Link]

PulseAudio is part of virtually all mainstream Linux distributions and by now it does tend to work pretty well. A dependency on Pulseaudio is unlikely to hurt the popularity of Firefox on Linux in any significant way simply because nearly every potential Firefox user runs PulseAudio already, possibly without even really noticing because it just does what it is supposed to do.

Presumably people like you who prefer ALSA without PulseAudio could band together and recompile Firefox such that it doesn't use PulseAudio (it's a compile-time switch). It is a bit of a hassle but at least you would be scratching your own itch. IMHO, insisting that everyone else ought to forego PulseAudio-by-default in Firefox just because you personally don't happen to like it is much more of an imposition.

Firefox 52.0

Posted Mar 14, 2017 7:46 UTC (Tue) by ceplm (subscriber, #41334) [Link] (1 responses)

> Even in theory it seems unlikely that adding another layer of bloat …

And dmix is just made of pixies, or what?

Firefox 52.0

Posted Mar 15, 2017 10:00 UTC (Wed) by cladisch (✭ supporter ✭, #50193) [Link]

> > Even in theory it seems unlikely that adding another layer of bloat …
>
> And dmix is just made of pixies, or what?

With dmix, all applications map the hardware buffer directly into their own address space, and mix the samples into the buffer with (the equivalent of) atomic adds. This avoid a separate sound server process/thread, and thus has 42% more pixiness.

And nothing is free: all applications must use the same sample rate/format and buffer size, so one fixed value must be chosen in advance.

Firefox 52.0

Posted Mar 9, 2017 21:22 UTC (Thu) by thestinger (guest, #91827) [Link] (52 responses)

It can be enabled at compile-time for now but it will be incompatible with the beginning of a sandbox landing in 54. It's an extremely short term solution.

Firefox 52.0

Posted Mar 10, 2017 0:02 UTC (Fri) by ajmacleod (guest, #1729) [Link] (47 responses)

If pulseaudio becomes a requirement then I'll be dumping Firefox the same way I did Skype after a very, very long time using it, so I'm grateful for the heads up.

Mozilla have long since used up all the good will I had towards them with their dogged determination to ignore what users actually want and their stupid focus on appearing to be a "progressive" organisation instead. I suppose I'll switch to Pale Moon, not being a huge fan of Chrome. Sticking to ESR releases has bought a little bit of time to digest the latest stupidities before I have to actually suffer them myself, but I'm pretty fed up of the whole cycle now.

Firefox 52.0

Posted Mar 10, 2017 2:56 UTC (Fri) by roc (subscriber, #30627) [Link] (45 responses)

Mozilla measured and about 1.2% of all desktop Linux Firefox users were using ALSA. So "what users actually want" maybe isn't what you think it is.

Firefox 52.0

Posted Mar 10, 2017 9:23 UTC (Fri) by jrigg (guest, #30848) [Link] (9 responses)

> Mozilla measured and about 1.2% of all desktop Linux Firefox users were using ALSA. So "what users actually want" maybe isn't what you think it is.

If most of the rest were already using Pulseaudio, why does anything need to change? Removing more options with each new release seems to be an ongoing obsession with the Firefox developers.

Firefox 52.0

Posted Mar 10, 2017 10:10 UTC (Fri) by cortana (subscriber, #24596) [Link] (1 responses)

Maintaining code isn't free.

Firefox 52.0

Posted Mar 10, 2017 16:40 UTC (Fri) by ajmacleod (guest, #1729) [Link]

True, maintaining code isn't free... at this rate though they won't have to worry as they soon won't have any code left that anyone actually uses! Perhaps it would be a better use of their resources to improve dialogue with their actual users rather than dictating and "moralising" about totally irrelevant issues.

Firefox 52.0

Posted Mar 10, 2017 19:46 UTC (Fri) by Cyberax (✭ supporter ✭, #52523) [Link] (5 responses)

Feel free to step up and maintain a fork with ALSA support.

Firefox 52.0

Posted Mar 10, 2017 21:34 UTC (Fri) by sbergman27 (guest, #10767) [Link] (3 responses)

Ah. The classic OSS cop out. Maintain it yourself. That argument would be more persuasive if Mozilla Corp were a non-profit and/or solely dependent upon volunteer efforts. But of course, Mozilla Corp is a for profit corporation from which over 1000 people get their paychecks, and with an annual revenue of well over $300 million. ($329 million in 2014.) Lets not forget that they went commercial in 2005 so that they could wheel and deal with the likes of Google and Yahoo, selling their influence (i.e. our views) to the highest bidder. There ought to be some money in all that revenue to continue the ALSA support. Especially since they cut Thunderbird loose because it was losing them money. Selling the position of default search engine in a mail client isn't nearly as lucrative as with a browser.

Firefox 52.0

Posted Mar 10, 2017 21:45 UTC (Fri) by tuna (guest, #44480) [Link]

You could sell views on you fork as well and make as much money as Mozilla does.

Firefox 52.0

Posted Mar 10, 2017 21:49 UTC (Fri) by pizza (subscriber, #46) [Link] (1 responses)

Okay, that's fine. Mozilla's a for-profit company. So they should focus on what's best for their customers.

Where exactly does maintaining ALSA support, targeting 2% of a userbase that consists of 2% of the market, fit into that?

What incremental revenue will they see for that ongoing development and testing effort?

Firefox 52.0

Posted Mar 14, 2017 7:50 UTC (Tue) by ceplm (subscriber, #41334) [Link]

2% of nice of their market … I am a big fan of Linux (working for Red Hat, after all), but I do not delude myself that it would be that significant part of the browser market.

Firefox 52.0

Posted Mar 10, 2017 22:14 UTC (Fri) by cortana (subscriber, #24596) [Link]

I don't think the backend is being removed from the source, so those who want to build Firefox themselves can stick with ALSA if they want.

Firefox 52.0

Posted Mar 17, 2017 12:00 UTC (Fri) by Darkmere (subscriber, #53695) [Link]

Because the Alsa backend doesn't work reliably in multi-process/thread access to the device.

Thus, right now, the alsa-backend, used by <2% of the Linux users, stands the choice of either put into it some serious developer effort, or disable it and be done with it.

Firefox 52.0

Posted Mar 10, 2017 16:36 UTC (Fri) by ajmacleod (guest, #1729) [Link] (15 responses)

And how did they measure that figure for pulseaudio/ALSA usage? Could it be that many users who actually know what they want (as opposed to those who just take whatever they are smacked in the face with) have as much telemetry disabled as possible?

In any case I wasn't really referring to this particular instance, rather a long, long trail of changes which nobody asked for or wanted (UI changes, "pocket", etc etc.) This will just be the straw that finally snaps the old camel's back in my case. Will Mozilla care? Clearly not. But they should, because it's my job to choose what software many other people use across a huge variety of businesses - and you can be sure their already plummeting marketshare will be dropping by a lot more than one user.

Firefox 52.0

Posted Mar 10, 2017 16:49 UTC (Fri) by jrigg (guest, #30848) [Link]

> Could it be that many users who actually know what they want (as opposed to those who just take whatever they are smacked in the face with) have as much telemetry disabled as possible?

This. I think it applies quite widely, not just to web browsers.

Firefox 52.0

Posted Mar 10, 2017 20:39 UTC (Fri) by sbergman27 (guest, #10767) [Link] (2 responses)

PA has always been a solution in search of a problem. And a highly unstable and problematic one, at that. On some machines I support, I still have to provide the user a desktop icon for killing and restarting PA to get their sound working when it fails.

Firefox 52.0

Posted Mar 10, 2017 22:20 UTC (Fri) by sfeam (subscriber, #2841) [Link]

That has been my experience as well. But it sometimes trashes its own configuration on failure. My first-level workaround is to get PA working once, then write-protect everything under ~user/.config/pulse so that the chance of being able to restart it is a bit higher.

Firefox 52.0

Posted Mar 21, 2017 16:13 UTC (Tue) by gerv (guest, #3376) [Link]

Useful... Can you comment and say exactly what commands that icon runs?

Firefox 52.0

Posted Mar 11, 2017 1:00 UTC (Sat) by roc (subscriber, #30627) [Link] (10 responses)

> Could it be that many users who actually know what they want (as opposed to those who just take whatever they are smacked in the face with) have as much telemetry disabled as possible?

Sure, it's possible that ALSA usage is as high as 2% or 3% of desktop Linux users.

> a long, long trail of changes which nobody asked for or wanted

Not "nobody", just "not me" and maybe "not some of the people I know".

The problem for Mozilla and other software developers is that a lot of users hate any kind of change and the only way to please them is to never change anything they'd notice. But if Mozilla stuck to that policy a whole lot of other users would mock Firefox for looking ancient.

Firefox 52.0

Posted Mar 11, 2017 1:03 UTC (Sat) by roc (subscriber, #30627) [Link] (9 responses)

(And to be clear, it would actually *be* ancient in important ways, and look out of place in modern desktops.)

Firefox 52.0

Posted Mar 11, 2017 11:39 UTC (Sat) by ajmacleod (guest, #1729) [Link] (8 responses)

This is a major problem in computing today - a disease, really. I deal with end users every day, and I can promise you that the vast majority of computer users DO NOT want constant flux and "modernising". If I had a pound for every time someone said to me in despair "what was that one a few years ago - Windows XP... that was the last one I could actually work" I'd be very wealthy. Notice that they don't just want the previous version!

It's not just a natural resistance to change, it's deeper - with Windows XP SP2 Microsoft had refined their GUI to a point where it worked well for the vast majority; nobody wanted ME or 98 back, even back then. Since then, MS went on a "modernising" spree that has left many, many people struggling to do with their PCs what they did before; sure it looks "modern" but essentially nobody wanted that.

With Open Source Software this kind of thing has also gone on - witness the Gnome 3 idiocy - but we've been free to laugh at it and carry on using any of the multitude of solid alternatives like WindowMaker. Now though, there's a distinct change of attitude / culture; this change at Mozilla, if true, is a prime example. I don't need PulseAudio on any of my machines and having had experience of it I won't touch it with a bargepole. PA is just another layer of bloat which is unnecessary for the majority of users (yes, it solves a few problems for a few people - great, feel free to use it) but having the flexibility to choose the makeup of our systems has been a major feature of OSS and culture for as long as I've known it and this is what's slowly but perceptibly getting more restricted.

Firefox 52.0

Posted Mar 11, 2017 13:29 UTC (Sat) by nix (subscriber, #2304) [Link] (6 responses)

PA is just another layer of bloat which is unnecessary for the majority of users

The majority of users (because the vast majority of machines sold) have Intel HDA-compatible (on-motherboard) sound. The vast majority of Intel HDA implementations cannot do hardware mixing, and if you try all but the first player will block until the sound device is released and you kill -9 it. Almost no players release the sound card unless you e.g. stop playback (not just pause it) and with a ramifying set of blocking processes like this it is very easy to be playing a fullscreen game and then the whole system appears to deadlock because the game has the sound card and something else has blocked on it and now the game is blocked waiting for some resource (often X-related) that the blocked app needs. Now you have to reboot, unless you have a second networked machine you can ssh in from and kill things. Oh and also because of the lack of proper cross-app volume control even if you *have* software mixing, the new thing playing its notification beep or whatever will probably be either inaudible or deafen you.

PA fixes all of that. But don't worry. The majority of users don't care if their whole desktop deadlocks because of some incoming-IM beep, or they get deafened by an incoming phone call!

Firefox 52.0

Posted Mar 11, 2017 19:49 UTC (Sat) by ajmacleod (guest, #1729) [Link] (5 responses)

My motherboard (and the one before that) also use Intel HDA - my system NEVER deadlocks due to multiple audio streams (or anything else come to think of it). I'm using ALSA - I didn't do anything clever at all to set it up and being on Gentoo I'd know all about it if complex configuration were required!

Audio works absolutely flawlessly on my system, recording and playback, and I am very happy to keep yet another pointless layer of flaky, poorly performing (by design) bloat out of the equation by banishing PA.

As I said before - I don't deny that PA offers features that some people need (and I needed myself on some systems in the past - the networked audio aspect); most people have no need of it at all, any more than most people need JACK.

I'm happy that PA is an option - I just want it to stay that way, an option.

Firefox 52.0

Posted Mar 11, 2017 22:32 UTC (Sat) by foom (subscriber, #14868) [Link]

The ALSA client library does mixing (dmix) by default these days.

Firefox 52.0

Posted Mar 13, 2017 18:20 UTC (Mon) by MattJD (subscriber, #91390) [Link] (3 responses)

>Audio works absolutely flawlessly on my system, recording and playback, and I am very happy to keep yet another pointless layer of flaky, poorly performing (by design) bloat out of the equation by banishing PA.

Out of curiosity, how is PA *designed* to be flaky or poorly performing? I can understand people believing it was *implemented* that way (let's avoid repeating that discussions here), but I haven't heard an argument about why it was designed that way.

Is it just an argument about PA being a userspace daemon vs doing the audio mixing in the kernel? Or is there something else?

Firefox 52.0

Posted Mar 14, 2017 3:07 UTC (Tue) by foom (subscriber, #14868) [Link] (2 responses)

People seem to have this misconception that ALSA is a nice simple kernel API and that pulseaudio is an unnecessary layer on top of the happy kernel-based alsa functionality they had previously been enjoying.

But it's really not like that. ALSA is confusingly the name both for low level kernel audio drivers, and some rather complex userspace library functionality, including sound mixing, samplerate conversion, multichannel remapping, and more. Effectively everything configurable it can do happens all in the alsa library, not the kernel.

Pulseaudio uses just the lower level kernel-alsa functionality and entirely replaces the higher level userspace-ALSA functionality (like mixing) with a new implementation. It provides a better API for apps, and also plugs in as the default sound output device in the high level alsa API, for compatibility with existing apps.

Firefox 52.0

Posted Mar 14, 2017 17:35 UTC (Tue) by mathstuf (subscriber, #69389) [Link] (1 responses)

> It provides a better API for apps

For evidence of this from an app developer, the comments and commit messages in mpv's direct ALSA support are a great source.

Firefox 52.0

Posted Apr 6, 2017 11:47 UTC (Thu) by nix (subscriber, #2304) [Link]

The mpv developer appears to be the new Jamie Zawinski insofar as comments that, ah, "tell it like it is" are concerned.

Firefox 52.0

Posted Mar 11, 2017 14:44 UTC (Sat) by pizza (subscriber, #46) [Link]

> with Windows XP SP2 Microsoft had refined their GUI to a point where it worked well for the vast majority; nobody wanted ME or 98 back, even back then.

Unless of course you had a system that couldn't run WinXP, due to inadequate resources (XP was considerably more demanding than 98 or ME) or lack of hardware support (2K/XP had very different driver models, and a lot of hardware never received drivers. (This was the golden era of "Linux supports your hardware better than Windows does")

> It's not just a natural resistance to change, it's deeper

No, it's absolutely a resistance to change. Nobody wants *anything* to change, ever. Except when they want something specific to change. And then they want it exactly the same, only different.

I learned this lesson twenty years ago, and it's a very large part of why I avoid UI and web-work like the plague.

Firefox 52.0

Posted Mar 10, 2017 16:59 UTC (Fri) by mgb (guest, #3226) [Link] (12 responses)

> Mozilla measured and about 1.2% of all desktop Linux Firefox users were using ALSA.

I'm responsible for linux boxen at a several facilities and none of them use pulseaudio.

Has Mozilla published the methodology it used to spy on users?

Firefox 52.0

Posted Mar 10, 2017 19:07 UTC (Fri) by pizza (subscriber, #46) [Link] (2 responses)

And I work at a facility with at least a thousand RHEL/CentOS end-user nodes, and PA is utilized on all of them.

In other words, the plural of anectdotes is not data.

Firefox 52.0

Posted Mar 10, 2017 20:30 UTC (Fri) by sbergman27 (guest, #10767) [Link] (1 responses)

What, exactly, is PA being used for on these 1000 end user nodes? What audio is it actually facilitating? Hmm?

Firefox 52.0

Posted Mar 10, 2017 20:46 UTC (Fri) by pizza (subscriber, #46) [Link]

They're full desktop sessions. As for what audio is actually used -- terminal bells, telecommunications/conferencing, teleconferencing, listening to music, watching a video on cnn.com, or whatever the user sitting in front of the session wants or needs.

The point is, PulseAudio is _widely_ deployed, to the point where the number of folks who don't have it (or aren't using it) are in the single-digit percentages. And, heh, even applications not using PA are probably going through an ALSA->PA wrapper behind the scenes.

Firefox 52.0

Posted Mar 11, 2017 1:01 UTC (Sat) by roc (subscriber, #30627) [Link] (8 responses)

Of course: https://wiki.mozilla.org/Telemetry/FAQ

The data is, of course, carefully anonymized.

Firefox 52.0

Posted Mar 11, 2017 1:51 UTC (Sat) by mgb (guest, #3226) [Link] (7 responses)

So Mozilla has no data from serious users then.

Firefox 52.0

Posted Mar 11, 2017 1:57 UTC (Sat) by rahulsundaram (subscriber, #21946) [Link] (6 responses)

> So Mozilla has no data from serious users then.

What does that mean, exactly?

Firefox 52.0

Posted Mar 11, 2017 2:02 UTC (Sat) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

Tinfoil-hat-wearing Veteran Unix Administrators. Duh.

Firefox 52.0

Posted Mar 11, 2017 10:07 UTC (Sat) by jrigg (guest, #30848) [Link] (1 responses)

>Tinfoil-hat-wearing Veteran Unix Administrators.

Does this kind of dismissive stereotyping contribute anything useful to the discussion?

Firefox 52.0

Posted Mar 11, 2017 13:33 UTC (Sat) by nix (subscriber, #2304) [Link]

It seems like a fairly accurate interpretation of a comment claiming that gathering telemetry by definition gains no data from 'serious users'. The implication I read was "users like mgb", but given mgb's comment history on this site (nearly 100% anti-systemd and anti-Poettering crusades and never subscribing as far back as I can remember) I don't see why anyone would feel any sort of loss at not gathering telemetry from a user like that.

Firefox 52.0

Posted Mar 11, 2017 2:15 UTC (Sat) by roc (subscriber, #30627) [Link] (2 responses)

I think it really means "me".

Firefox 52.0

Posted Mar 11, 2017 13:35 UTC (Sat) by nix (subscriber, #2304) [Link] (1 responses)

Quite. Turning telemetry off and then whining that Mozilla broke features you use seems profoundly self-defeating to me. The point of the telemetry is to tell that the features you use are actually in use: by turning it off, you're saying that you don't care if Mozilla doesn't take your personal usage patterns into consideration by making changes. If it then goes ahead and doesn't take your personal usage patterns into consideration, it is just doing what you asked!

Firefox 52.0

Posted Mar 11, 2017 19:37 UTC (Sat) by rompel (guest, #4512) [Link]

Not really a fair argument given that telemetry is off by default in release builds.

In fact, with telemetry only coming from users who have installed a pre-release build (and not opted-out) or have manually turned on an obscure preference knob, it's hard to make the case that the data comes from a representative sample of users.

Firefox 52.0

Posted Mar 10, 2017 20:42 UTC (Fri) by MoSal (guest, #103113) [Link] (2 responses)

And what is the percentage of desktop Linux users submitting telemetry data? Can you even know that number?

Another ESR/ALSA/Vimperator user here, with an overridden user-agent btw.

Firefox 52.0

Posted Mar 10, 2017 22:39 UTC (Fri) by khim (subscriber, #9252) [Link] (1 responses)

And what is the percentage of desktop Linux users submitting telemetry data? Can you even know that number?

Why would I even want to know that data?

As with any democracy voters who actually arrive at ballots decide the outcome. If you disable the telemetry - you are basically saying "my opinion does not matter, please ignore me". That's fine, that's Ok, but then - please don't complain if decision would be done against your wishes.

Firefox 52.0

Posted Mar 10, 2017 23:08 UTC (Fri) by MoSal (guest, #103113) [Link]

That's a lovely false equivalency you got there. But not as lovely as
the almost-cute belief that elections are *that* relevant.

I will stop here before wandering further off topic.

Firefox 52.0

Posted Mar 10, 2017 23:24 UTC (Fri) by sbergman27 (guest, #10767) [Link] (2 responses)

Hmm. 1.2% is about the same as the percentage of total Firefox instances running on a Linux platform. Perhaps it's time for Mozilla Corp to drop Firefox support for that niche OS? Almost nobody uses it on the desktop, after all.

Firefox 52.0

Posted Mar 11, 2017 0:00 UTC (Sat) by khim (subscriber, #9252) [Link]

This maybe good long-term solution, but short-term this would incur huge costs because lots of Mozilla developers use Linux. They are basically maintaining it for in-house use and other Linux users get it I for free because of that. Raw ALSA, on the other hand, is not used by Mozilla developers, presumably, thus it's support is pure waste.

Firefox 52.0

Posted Mar 11, 2017 1:06 UTC (Sat) by roc (subscriber, #30627) [Link]

2% of all Firefox users is still a pretty big number.

2% of that 2% is not such a big number. Also, dropping ALSA support does not make it impossible for those users to use Firefox, like dropping desktop Linux support would.

Plus what khim said; there are real advantages to doing Firefox development on Linux. At least a few Mozilla developers are on Linux so they can use http://rr-project.org more easily...

Firefox 52.0

Posted Mar 11, 2017 10:39 UTC (Sat) by bluss (guest, #47454) [Link]

Firefox devs are prone to saying that Windows is their most important platform by far. (Given their user numbers).

It's unfortunate that they have forgotten their open source ideals/alliance with Linux to that degree.

Firefox 52.0

Posted Mar 11, 2017 9:58 UTC (Sat) by iive (guest, #59638) [Link] (3 responses)

> Posted Mar 9, 2017 21:22 UTC (Thu) by thestinger (subscriber, #91827)
> It can be enabled at compile-time for now but it will be incompatible
> with the beginning of a sandbox landing in 54. It's an extremely short term solution.

The sandboxing of ALSA has already been fixed:
https://bugzilla.mozilla.org/show_bug.cgi?id=1309098#c18

Do you know of some other fundamental issue that cannot be easily fixed?

Firefox 52.0

Posted Mar 14, 2017 14:53 UTC (Tue) by glaubitz (subscriber, #96452) [Link] (2 responses)

> The sandboxing of ALSA has already been fixed:
> https://bugzilla.mozilla.org/show_bug.cgi?id=1309098#c18

That's not a fix. It's a temporary workaround. They are opening the sandbox, thus increasing the attack vector.

Firefox 52.0

Posted Mar 14, 2017 15:45 UTC (Tue) by pizza (subscriber, #46) [Link] (1 responses)

Okay, I'm missing something here.

How is this an increase of the attack surface vs the former status quo of no sandboxing whatsoever?

In order to play audio, you need access to audio devices, yes?

Firefox 52.0

Posted Mar 14, 2017 15:51 UTC (Tue) by glaubitz (subscriber, #96452) [Link]

> How is this an increase of the attack surface vs the former status quo of no sandboxing whatsoever?

Sandboxing is already enabled, albeit not for all components and environments yet:

> https://wiki.mozilla.org/Security/Sandbox

> In order to play audio, you need access to audio devices, yes?

Or have something like PulseAudio which accesses the audio devices for you without breaking the sandbox.