On Fri, 2008-05-30 at 16:49 -0400, Alan Cox wrote:
Why the ... are people still writing software which doesn't try
tolerate faults that are recoverable to a useful extent. Yes dbus
might have to lose a few messages and send everyone a "duh whoops"
event so they can recover but "oh dear it broke everyone reboot" is
not good engineering.
This is exactly the problem that we were discussing some time ago
about PulseAudio. It is a great daemon, but right now it crashes
reliably for me, and until I figured out that I needed to run
"pulseaudio -D" from a terminal, my only way out was a reboot!
This is pretty brutal. I understand that we aim for perfection,
but we're far from it. The thing crashes reliably here after
a few hours of use, and there is No Way(TM) you can expect regular
human beings to issue a "pulseaudio -D".
When you put 2+2 together it follows that we have a desktop
that normal users must completely restart every few hours to
get working sound!
When I brought this up last time, I was told by the developers
that it's normal behaviour for most software out there, and that
I should file bugs and work toward marking PA perfect.
I don't understand why it's so inconceivable for PA to restart
itself? There isn't all that much state to begin with, and _any_
user would prefer a sound glitch to a full desktop restart
(or even to a "pulseaudio -D"!).
Dimi Paun <dimi(a)lattica.com>