[asterisk-dev] Pthread wrapper updates

[SF Markus Elfring]

> I strongly oppose the introduction of this approach.  As I noted before,
> I can't imagine any case where aborting the application is the right
> thing to do in the case that some sort of minor error condition is
> detected.  We should always do our best to deal with error conditions
> so that the application continues.
>   

It depends also on the used means and techniques for fault-tolerance. 
How do you think about experiences from some research areas?
http://en.wikipedia.org/wiki/Crash-only_software
http://en.wikipedia.org/wiki/Fault-tolerant_system
http://en.wikipedia.org/wiki/Load_balancing_%28computing%29

How often will a safe restart or quick (micro-)reboot be required to 
guarantee consistency and correct program behaviour for a while?

Regards,
Markus