Glitch-Free PulseAudio in Rawhide

[Lennart Poettering]

On Wed, 21.05.08 09:25, Dimi Paun (dimi at lattica.com) wrote:

> It is completely irrelevant that this is exactly what happens.
> In fact, you could say that "a number of bytes don't reach a process
> via file descriptor 7" and it would be exactly what happens, all
> at the same time being a cryptic and bad error message.
> 
> Do you really think a non technical user knows what a "server" is,
> especially in this context? If you are tempted to reply yes, don't
> bother, it Just Plain Wrong (TM).

Have you noticed that most error messages on Linux are this terse? I
think we simply disagree here what the main focus here should be. I
think the focus should be to fix the remaining issue for you so that
the entire error can go away. Your focus seems to be to make failing
more descriptive. -- I am not at all against making error
messages more discriptive, I would happily accept a sensible patch for
this. So, please stop complaining about this, get off your ass and put
your code where your mouth is! 

But please, strerror()-like error messages is how error reporting is
mostly done on Linux these days. I am not sure why you are picking on
PA that much. In contrast to what you claim desktop audio is really
not a core component of the system. If audio breaks the effect is
rather, huh, irrelevant most of the times. I can list you a lot of
components of a modern system where failure is far, far more
catastrophic. And no, I am not using this as an excuse that I can
produce low quality, buggy code. All I want to ask you: please keep
things in relation. So you are experiencing one bug on your specific
setup that makes listening to your favourite music break after a few
hours of continious play. Surely annoying, something we need to
fix. But just one bug, and not catastrophic. And you are raving how
really really essential this is, something that "MUST JUST WORK!", and
blah and foo. And then you claim we ignored systematically and was
just one among many. But that's not true. Lubomir responded. As did
I. Maybe we didn't treat your bug with the priority you'd have liked
that we treated it with, sure. But really, no need to flame anyone about.

Please, let's stop this flame war here now. Bugs happen. All the
time. Bugs aren't discovered by QA, quite often. This one wasn't fixed
before F9 release. 

> Not by a long shot. If an app crashes, _anybody_ knows to restart it. On
> this other hand, sound is a system feature that must work. The fact that
> it doesn't without frequent reboots reflects a design problem -- sound
> must restart automagically, not require the user to reboot every hour or
> so to recover this essential desktop feature!
> 
> For fsck's sake, I didn't have to reboot Windows 3.1 that often!

This is nonsense. And you know it.

Lennart

-- 
Lennart Poettering                        Red Hat, Inc.
lennart [at] poettering [dot] net         ICQ# 11060553
http://0pointer.net/lennart/           GnuPG 0x1A015CC4