Monday, January 26, 2009

Give Lennart a break

Even though Lennart has been working hard to make people lives easier, it seems some people are working hard to make his life harder by making discouraging comments not just about his ideas but the great work he has been putting up.

First of all, if you don't know why a sound server is needed on a modern desktop environment, you need to research about it (asking Lennart or anyone who knows *politely* might be a good start). Making a big fuzz of your ignorance and being proud of it won't help anyone, especially yourself. If you had asked this question from Lennart and he didn't give you a nice satisfactory answer at that time, it must have been because he must be getting really tired of answering this same exact question so many times, over and over again. That doesn't mean a sound server is not needed. No need to rush to conclusions. Ask someone else and/or research more. God is in the details.

Secondly, if Lennart's software breaks some proprietry shit, it's really not his fault and the last thing we need is some random mono-hacker who doesn't even know why a sound server is needed, to blog about Lennart being the guilty party.

Thirdly, there is nothing wrong with GNOME having a hard dependency on PA and there is no need to uninstall GNOME. This is as silly as saying "so if you don't want to have gtk installed on your machine, just stop using GUI apps or change desktop".

No really WTF do you guys want from him?


David said...

I agree people shouldn't be mad at Lennart.

But I do quite sympathize with people who want to be mad at someone over these things. Really, they should be peeved at their distro, for switching before the bugs were worked out.

(Or, at least, they should be if their distro is like Ubuntu and says one of their goals is user-friendliness.)

Russ said...

I'm sorry, whether sound works the way users expect is very visible. Someone can pretend they live in some ideal world where users don't depend on proprietary software, or where everyone has an egalitarian sound setup, but if all developers thought that way, nothing would work. Part of developing software is realizing that (a) your software has bugs and shortcomings and (b) other software has bugs and shortcomings. You can't just cut corners on your own software and ask everyone else to add features to their software so that yours works properly.

Anonymous said...

I agree, if you don't like how something works or if it has issues. try to help, by reporting useful bugs and maybe contribute patches.

Bitching about it will not improve it, starting another project to solve the same problem is not the answer either, you wind up with a lot of duplicate effort and 2 projects.

Anonymous said...

Irony: when a hacker working on *mono* blogs about having to remove software to make his desktop work.

user said...

I personally don't want anything from him, but I consider that shipping software that's not working without problems on such a huge computer base in a stable Linux distributions is bad. It's bad for users, for the users that don't even know Lennart, for the users that just want their Linux system to WORK!.

Joe said...

While Flash may well be "proprietary shit", its availability on Linux is a key factor in my family's willingness to run Linux. Same goes for a whole lot of users. So, unfortunately, any patch that breaks Flash has to be considered a bug that would prevent the use of the software in most distros.

oliver said...

If you are unhappy because Pulseaudio broke sound on your distribution: blame the distro, no the upstream devs. In the end the distros decided that PA was good enough for inclusion.

Anyway, even though there are problems now, I'm sure they will be sorted out, and in the end we'll have a much better audio system. I mean, ALSA is nice so far for getting sound in and out, but I still have to fiddle with five different mixer controls to enable the microphone on my SB Live - that's pretty sad. Looks like PA will fix lots of such problems.

Kieran Clancy said...

Pulseaudio would be great if it worked reliably, but for me at least that is rarely the case. It consistently hangs, freezes, locks up, hogs CPU and skips (underruns) on my machine. It is utterly frustrating and there seems to be no easy way to go back to plain ALSA (on Fedora 10). For me this has nothing to do with proprietary applications -- I get all these problems with audacious, totem, etc.

I use my machine as an alarm clock, and since Fedora has moved to Pulseaudio I can no longer trust it to work properly -- I have to set another alarm as a backup. On several occasions where I haven't set another alarm for a backup, the audio has failed to work (or has only come on for a fraction of a second before hanging) and I have been late for work.

Recently I tried to get a microphone working and after an hour of trying had to use Windows instead. The volume levels would reset when you looked at the computer the wrong way, and somehow Pulseaudio has made all my audio and especially my microphone levels very quiet. I used to be able to record using my microphone on this machine without any problems at all.

I'm not mad at anyone about this, but I can say that it has been an extremely frustrating experience for me ever since my distro went to Pulseaudio.

Anonymous said...

Thank you, my sentiments exactly. Lennart is a true hero, and he mostly gets crap for it by people who just have no clue. Being a mono hacker is just one extra symptom of being that clueless.

Bob said...

A patch that breaks "proprietary shit," in your words, is still breaking something whether you personally care about it or not.

It is this sort of attitude that puts a damper on Linux adoption, remember most people do not share your ideals nor do they want to, they just want to watch amusing flash videos with perfect audio.

Vadim Peretokin said...

Flash is important to me too, and PulseAudio integration so far has been very poor. Why should a user be happy they got more troubles?

Jeffrey Stedfast said...

Not everyone has the time or desire to spend time on fixing the bugs in a sound server that they can't easily replace if it's broken.

If PA was a normal application that ran on the desktop, I'd be more inclined to agree with your point of view, but it's not.

I also was not mad at Lennart, I was mad at my sound not working and completely breaking my entire system (e.g. entire desktop locking up, unable to login or logout, etc). If it was as simple as sound not working, it would have been less frustrating.

That said, my PA rant was like 8 months ago. Old news, especially since I already did as you suggested and more:

- I reported the bugs.

- I worked with the developers to fix them.

- I fixed a number of them myself.

Rowan Lewis said...

Joe, be realistic, if we didn't start using PA before flash was fixed to work with it, flash would never get fixed at all. Progress isn't always entirely positive.

will_in_wi said...

I think the primary reason for all of this frustration is that many (most?) linux desktop fans don't see the reason for pulse. KDE4/3 works great without it, Gnome worked great without it, everything user visible just worked. Pulse added another layer of abstraction and bugs, as most new software does (see KDE4). It got adopted perhaps a little too quickly by the distros.

I personally have not seen a compelling reason to use pulse. Alsa's dmix seems to do all of the important stuff with much less overhead. The main reason to use pulse seems to be the ability to switch output streams between devices on the fly and have sounds be aware of screen position and other activities going on (background music pausing when a call comes in, etc). This is a really cool feature, but has almost nothing to do with most user's workflow. Perhaps this would be better put into alsa proper. I have seen many of the features in pulse done through asoundrc, although in a insanely complicated way. Perhaps some patches to add the missing features to alsa, plus a nicer interface is the right way to go.

I think that gnome should have a hard dependency on gstreamer, and then let gstreamer take care of the sound server. I thought that everyone was trying to get away from the hard dependency on a sound server due to being burned by arts and esd. KDE's response was phonon and Gnome's response was gstreamer. A new sound server comes around and suddenly everyone must use it. We have stable tested abstraction layers to allow testing and phasing in of new tech with the minimal user disruption. Why aren't we using it?

There is way too much anger and flaming on both sides of this issue. Many anti-pulse people appear to be trying to figure out why this piece of software is causing so much trouble and responding badly. Many pro-pulse people appear to be assuming that everyone wants a super-smart sound manager, when I think the reality is that most people don't care about this part of the desktop as long as it works. Could this discussion become a little more civil and reasoned?

lat said...

Well, when something that doesn't work properly is forced into people like that, they are going to complain, and not everybody has time to ask the developer (especially if he's not likely to answer because he's "really tired").

Remember what happened to the "other desktop environment" after most of the major distros forced upgrading to the new version, even though half the developers advised against it.

So yeah, you're right, but that won't save Lennart from getting flamed unless the distributors drop Pulse altogether until it's fixed. And I'm sure he doesn't want that either, so complaining about the flames makes little sense anyway.

jbr said...

On my (fairly common) hardware, pulseaudio doesn't work well with voip apps such as ekiga and empathy. Playback is very choppy and the microphone only records at very low volume (no, it's not a mixer setting). Reverting to plain alsa fixes all these problems.

Until these bugs are fixed, pulseaudio is simply not an alternative for me. In my opinion, it's way too early to make gnome depend on pulseaudio.

Anonymous said...

Right, because having working flash isn't important at all... oh wait.

Actually, it's pretty easy to argue that working flash is more important than having a sound server at all (and I do know the reasons for having one).

James Henstridge said...

Jeff ran into real problems with PulseAudio and how it was configured on SUSE. As well as bitching about them, he provided a number of patches to fix his problems.

If someone was going to complain about my software being broken, there are worse ways to do so.

As for what people want from Lennart, it'd be something close to feature parity with the old system (this doesn't necessarily mean that the features need to be exposed in exactly the same way though).

Glitchless audio and per-app volume controls are nice, but I'd prefer that the basics were working correctly first.

Despite this, I do understand that the security blanket needs to be removed at some point or people will use the fallback mode and never test the new code.

With the recent changes, we are betting the farm that the existing PA adoption blockers will be fixed for the next GNOME release. If they aren't, distros are left with the decision on whether to disable the new PA/GNOME code or roll back to the equivalent 2.24 code.

fogman said...

Sorry to say that, but PulseAudio is breaking up my system all the time. I randomly have sound (hardly), even for GStreamer apps. Not to mention the problems which PulseAudio is causing with proprietary software. If i want to have (perfect?) Sound, i'd better kill PulseAudio.

I never complained about these issues, because i know that they will be fixed soon as possible.
But i really don't like, if developers (me too) don't listen to other peoples complains and even worse, making those people look like idiots on the net.

Especially, that those people are contributing software to the open source community and even patches to some (in this case PulseAudio) problems.

Sorry! I think you have gone to far.

thomas11 said...

"WTF" we guys want from Lennart? Working sound on our machines. You can tell me all you want about how a sound server is needed, fact is that there is a quite a big group of people for whom sound worked at least satisfactorily before PA, and now it doesn't anymore. A quick google search immediately shows the multitude of problems.

Sure, if the fault lies with another piece of software it's not Lennart's fault but the distributions. But no one can tell the thousands of frustrated users that PA is mature. And many are *not* convinced that it's even needed.

Anonymous said...


My Adobe flash works just f-ing fine on my laptop AND my desktop systems.

Both are running Fedora 10. Both have used the Intel-HDA stuff at the beginning now with my desktop I have a nice M-Audio 24/96 audio card stuffed in it.

Also Wine plays fine. WoW for the win.

Although Ekiga gives me fits.

All of this works perfectly fine using Pulseaudio.

The only tweaks I had to do was add myself to the Pulse-RT and enable higher priority stuff in the pulseaudio configuration file.


So don't blame PA when your distribution is the one that f*ked up PA.

And, yes, I am pointing fingers at Ubuntu. I know that Fedora still has some issues, but the majority of crappiness is coming from Ubuntu.


Of course nothing is perfect. The Fedora 10 Intel graphics drivers just blow donkey balls.


PA kicks ass.

Try this without using PA... Plug in a USB audio device that is supported by Alsa.

Start playing music on your onboard audio device.

Now migrate the audio from the onboard device to USB.

Now once that is done play using both devices at the same time.

Now that is done start transmitting your audio over the network to ANOTHER device.

That sort of thing takes me, ohhh.., about 10 seconds of clicking in PA's volume control.

Anonymous said...

I'm not mad, personally I like the idea of pulseaudio and I'm interested in some of its feature (mainly the per-program volume control). My problem is that I've got a soundcard with hardware mixing, which works well, and I don't have the time to track down bugs and solve problems.
So I'm not going to switch anytime soon, and I'm PISSED that the switch is being forced on me, because the old functionalities are being removed even though pulseaudio is all but pervasively adopted.

Most of all: why a f*cking notification icon? why do you want to break my personalized layout? can't you just leave the old volume control applet available, or make the new pulseaudio-based volume control a traditional applet?

Anonymous said...

From what I remember about Jeffs post he actually put his code where his mouth is and solved the problems he found...

Anonymous said...

You're completely missing the point. If you can't understand why people are angry when something that used to work stops working, ask yourself if you care about users. The problem is not about PA, nor Lennart. The problem is using the newest stuff even if it means regression for the end user, and annoying our user base.

Livio said...

I think you should "rethink" what you said.

PulseAudio pre-0.9.10 worked well, at least for me. No hiccups, crashes, slowdowns etc. PA has been broken. While I haven't been complaining, I started to do so.

I hate insulting and complaining all the time but I can't stop while I'm using new PulseAudio. The worst thing about it is that PA is still being made ready for desktop, while it has been already put into stable versions of distros.

This was done too fast and with big disadvantage for us.

I hope he'll fix the stupid issue he (and maybe others) introduced and say "kiss my ass" instead of fixing them.

What's sad, we can't just remove PA, because it's currently going to be GNOME's dependency...

Bad, bad, bad! Too many reasons to be mad.

maninalift said...

(1) PulseAudio isn't gtk, if you use it then you use it for all your apps (or as many apps as will work at all). It just isn't the same thing.
(2) Saying that people are idiots if they don't understand why they need a sound server doesn't help. There may be good technical reasons for it but the user knows his experience and that should not be ignored.
(3) I don't think users' primary concern is whether their audio is a bit hit-and-miss for a few months: it's whether free-desktop audio will be in a better state afterwards.

Anonymous said...

Wikipedia tells me nothing of why I would want a sound server on my desktop.

All I want from my desktop sound system is *per application or per stream volume controls*.

I've used GNOME on my machine for 3 years now without a sound server with no real ill effects or want for more.

zeenix said...

I've used GNOME on my machine for 3 years now without a sound server with no real ill effects or want for more

This and other similar comments remind me of the discussion I once had with my father:

Me: I need a computer.
Father: Why?
Me: because of blah blah and blah (pardon me for not remember my arguments at that time)
Father: oh? but if it was so damn important thing, how come I never ever needed it and have lived a happy life without it?
Me: (sigh) :(

Anonymous said...

Why did you bring up that Jeff was a Mono hacker? He has contributed far more to GNOME than he has to Mono, and far more to GNOME than you have.

Shouldn't you have said "gnome hacker"?

zeenix said...

Shouldn't you have said "gnome hacker"?

Dude! have you looked at his blog/website? There is a big label on the top, claiming his contribution to mono but I don't see anything similar about GNOME. There is some links/tabels of GNOME but for each time GNOME is mentioned anywhere, mono is mentioned twice.

Maybe he has contributed much more to GNOME than I ever will but that is besides the point. He seems to be wanting to be known for his mono contributions more and that is exactly what i know him for.

Anonymous said...

The fact that he has contributed to Mono is completely irrelevant to his PA rant, so I'm left wondering why you brought it up at all.

A little digging later and I find that you're the author of GUPnP which has taken you a long time to implement in C/Vala.

Meanwhile, the Banshee author wrote a UPnP stack from scratch in a less than a week in Mono.

I think that's the source of your anti-Mono angst and that's why you brought up the fact that Jeff is a Mono contributor. You needed to make yourself feel better by slighting Mono and Mono developers.

oliver said...

Btw. if your microphone volume is too low with PA, try installing padevchooser (and the associated pavolumecontrol or what's it called), and then you might find a separate mic volume switch in pavolumecontrol which is at 30% or something... Setting this to 100% gave me full volume on my Hardy system :-) and I hope this will be fixed properly with the upcoming new Gnome volume control.

zeenix said...

A little digging later and I find that you're the author of GUPnP which has taken you a long time to implement in C/Vala.

Meanwhile, the Banshee author wrote a UPnP stack from scratch in a less than a week in Mono.

You might want to dig a bit more to get the facts right:

1. Jorn is the main developer of GUPnP project but he was mostly busy with his studies during this whole time.

2. I was mostly working on it using bits and pieces of my spare-time until very recently that i started to get paid for working on Rygel (MediaServer on top of GUPnP). You can check the commit log on Rygel to see how fast it has been developed since then.

3. GUPnP project has much more to offer and addresses a much bigger audience than the banshee plugin you are talking about. Does banshee plugin offer developer tools for UPnP like we do? Is is possible to use this stack standalone and if so how can one possibly use it outside the context of Mono? Does it provide a generic, simple yet very powerful API like GUPnP does? e.g is it possible to implement a NetworkList or IGD on top of it? Is there nice API docs somewhere I can browse through? I can go on on this one if you like me to but i think you get the point. :)

4. Jorn wrote a similar rhythmbox plugin on top of GUPnP and that took him a day or so in total.

I think that's the source of your anti-Mono angst and that's why you brought up the fact that Jeff is a Mono contributor.

Heh! You can think whatever you like. I can only smile and tell you that this is not the case.

I will admit that i was disappointed when Banshee guys wrote their own UPnP stack instead of using GUPnP but that disappointment (not anger) was based on the fact that they gave me clear indication that GUPnP is the way to go forward on UPnP front at GUADEC. However, this is thing of the past and i don't hold any grudges against them for making this decision. To each his own.

Ian Hurst said...

Pulse Audio is fail for me. Crashes regularly, especially when browsing. Before the switch this didn't happen. Yes, I know, the old way had its own issues, but those issues were less trouble than the constant hang ups.

I'm sure Pulse Audio will be great in time. I know it was built for the right reasons. Its authors should be thanked. If you're looking for a scapegoat, direct your rage at the distributions that included it before it was stable. Nobody forced them to switch so soon. I for one have an axe to grind with Ubuntu now...

Anonymous said...

However, this is thing of the past

8 months ago was the past, but it didn't seem to stop you from attacking this guy.

If people are still bashing PA 8 months after that guy's rant, maybe he was right; maybe it's time for the PA developers (Lennart included) to get off their high horses and start fixing their shit.

People are still having problems with PA on Fedora 10 which came out only a few months ago and you can't put the blame on anyone but Lennart there because he's the packager AND the developer.

zeenix said...

8 months ago was the past, but it didn't seem to stop you from attacking this guy.

Which guy? If you are talking about Jeff, in case you haven't noticed he wasn't involved in there at all. Also, you seem to have missed that this blog entry is more about defending a particular person and attacking.

zeenix said...

and attacking.

I meant "than" attacking.

Anonymous said...

How is Jeff not involved? You linked to his blog attacking him for ranting about PA 8 months ago.

zeenix said...

How is Jeff not involved? You linked to his blog attacking him for ranting about PA 8 months ago.

You are mixing things up. Let me try to explain:

However, this is thing of the past

I said this in the context of the totally irrelevant GUPnP issue that you brought into the picture. You said in response to this sentence:

8 months ago was the past, but it didn't seem to stop you from attacking this guy.

So yes, Jeff is not involved in that issue (if there is one in the first place) at all.

Anonymous said...

I think you are confusing the multiple Anonymous posts =)

I didn't mention anything about GUPnP, that was a different Anonymous.

My point was that you attacked Jeff for a rant he posted 8 months ago and brought in an irrelevant fact about him contributing to Mono for no apparent reason other than to attack Mono?

You also go on about how things are in the past blah blah, but you attacked this guy for something he said 8 months ago!?

Things just don't add up.

Vadi said...

"People are still having problems with PA on Fedora 10 [...] he's the packager AND the developer."

Ouch, and here I thought all blame was on distros for "not implementing PA right".

Anonymous said...

This and other similar comments remind me of the discussion I once had with my father:

Me: I need a computer.
Father: Why?
Me: because of blah blah and blah (pardon me for not remember my arguments at that time)
Father: oh? but if it was so damn important thing, how come I never ever needed it and have lived a happy life without it?
Me: (sigh) :(

See, the difference between that and this would be if your father got you a new computer, and it was constantly blowing the circuit in the house, so now his lights, TV, and other devices have become unreliable. It doesn't matter that the computer (sound server) itself may be useful, if it doesn't work correctly and breaks several other things that were previously working, its not going to be wanted. I work with Linux, I deal with it 12+ hours a day, I have to use Ubuntu, Fedora, RHEL, and other distros at work on a daily basis. Every single one that uses PulseAudio, all on different hardware (I've tested at least 9 different machines) has serious problems with Pulse Audio. I'm really wondering if ANY machine works correctly with it, cause I have a very wide variety of hardware.

The right path to take is more similar to NetworkManager, which worked around many common bugs, was easy to disable, and allowed us to have a smooth transition until it finally become mature and stable enough to be the default network stack. Taking the shortcut of not having workarounds for common driver bugs, randomly failing, and providing no way to get online when it does would have made users hate it too.

master said...

Very nice
I write a blog about stoping the fixed microphone
and i tested also in linux
This is the link
thank you and nice work

Anonymouse said...

Clearly JACK is a better candidate for system-wide sound server, as it easily works well with apps that depend on ALSA or OSS (which is pretty much everything)

Fact is, the new kid in town thinks he's tougher than he really is.