Glitch-Free PulseAudio in Rawhide

Callum Lerwick seg at haxxed.com
Tue May 20 23:00:10 UTC 2008


On Tue, May 20, 2008 at 4:52 AM, Lennart Poettering <mzerqung at 0pointer.de>
wrote:

> And just respawning when we crash is taping over bugs, too. I am not
> generally opposed to it, though.


I think you need to meditate upon the Rule of Repair for a while:

 http://catb.org/~esr/writings/taoup/html/ch01s06.html#id2878538

Bugs happen. The manner in which you recover from them is the difference
between getting useful bug reports and getting flamed to hell and back on
every mailing list and blog in existence and making the entire distribution
look bad.

And there's the matter of appropriateness. Forcing users to stop and deal
with bugs makes sense in, say, rawhide. People should not be *forced* to
play QA in a stable release. They just want to Get Their Shit Done.


> However I fear it won't help much
> since PA's state is lost and thus all music would stop playing
> anyway.


This situation seems entirely recoverable to me. Why can't the clients wait
around and simply reconnect to the daemon and continue when it comes back
again? Sure, there would be a temporary glitch. Temporary is good. A
temporary glitch is far preferable to permanent brokenness, like things just
no longer working with no indication of what happened or how to proceed to
fix it. A temporary glitch is ignorable. Yet it is still a glitch, so a user
with time on their hands can make the *choice* to stop and investigate
further. Taking away that choice takes away the user's control. Taking
control away from the user makes the user unhappy. Unhappy users go out on
mailing lists and blogs and endlessly flame you.

To wit:

http://www.joelonsoftware.com/uibook/chapters/fog0000000057.html
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listman.redhat.com/archives/fedora-devel-list/attachments/20080520/b2184915/attachment.htm>


More information about the fedora-devel-list mailing list