On Sun, 2009-02-01 at 22:53 +0100, Martin Olsson wrote: > PS. I think Lennart is doing a _terrific_ job; I'm hoping Ubuntu technical > board understands the need to be careful about merging new stuff to avoid > regressions. This experience has been quiet painful for me and I suspect > there is other people still out there with PA related regressions. DS. > It's not that simple, in fact I'd go as far to say that "we should never adopt new things" is a very dangerous position to take.
Ubuntu gained its initial reputation by being cutting edge; we were one of the first distributions to truly embrace Linux 2.6, to base our distribution around hotplug (and later, udev), to make user mounting of pluggable devices "just work", to enable compositing, to enable easy wireless management, etc. If we were to stop now, and declare that Ubuntu won't embrace new technologies or software for fear of regressions, Ubuntu in 10 years time will look like Ubuntu today -- except that a newer distribution will be cutting edge, and have the users. When we enabled compositing in our X server, we found that most of the drivers were broken in some way. But this didn't stay true forever. By enabling compositing, the problems with the drivers were very visible, bug reports were received - and drivers have improved across the board. When Network Manager was installed by default, we found that most of the wireless drivers were broken. Now we have a new wireless stack, and for a lot of the time, they're pretty bullet-proof. We've enabled PulseAudio, and now we've found that most of the audio drivers are broken. I strongly believe that they will be rapidly fixed now that bug reports are flowing in. I also strongly believe that if PulseAudio were turned off again, the flow of bug reports would be stopped and that the drivers would not be fixed. (After all, they wouldn't be broken if PA didn't have a catalyst effect). A better question would be to ask what features PulseAudio provides[0], and whether they are interesting for us as a distribution. It provides: - mixing of multiple sound sources, - to multiple sound outputs, - at different volume levels, - using DMA The first of these is an obvious one; you want to be able to play music, get sound events and hear your VoIP phone ringing. There are other pieces of software that do this already. The second of these is a major feature of PA. It supports multiple output devices, this is something that dmix, esd, etc. do not do. Simply put, this means your Bluetooth headset works *and* you can make a VoIP call on that while your sound card is still playing music. The third is another major PA feature, the volume control of each sound source can be controlled independently. When your VoIP phone rings, the music volume can be decreased, when you answer the call - it can be transferred to the bluetooth headset you just turned on, and the music volume returned to normal. (If you don't pause it). The fourth is part of the "glitch free" PA work. Simply put, it means that large sound samples are queued for your card using DMA, rather than interrupts and a sound buffer. Fundamentally, your processor can sleep while your music is playing. So PulseAudio gives us support for simultaneous sound from multiple applications, support for headsets and other output devices, support for "user friendly" manipulation of the volume, and improved power management while playing audio. This comes at a cost of finding bugs in the audio drivers which should/will be fixed anyway as time goes on. Scott [0] another example here is Tracker; which we deployed and subsequently removed. The removal was because it was clear that users were not benefiting from it, thus the cost was not warranted. -- Scott James Remnant sc...@canonical.com
signature.asc
Description: This is a digitally signed message part
-- Ubuntu-devel-discuss mailing list Ubuntu-devel-discuss@lists.ubuntu.com Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel-discuss
