Hi,

on Fri, 28 Apr 1995 14:18:33 +0200 Paul Kranenburg wrote 
concerning "sleep(3)" something like this:

> Currently, sleep(3) and usleep(3) never return if the process' set of blocked
> signals happens to include SIGALRM. I'm inclined to believe that this 
> shouldn't be so; [u]sleep() sets ist own ALRM handler and can therefore
> enable SIGALRM when sigsuspend() is called.

Although I can't think of a case, where this would be critical, I'd take
it to be a bug, if usleep does not return after the specified amount of
time. And it's quite easy to fix anyway.

Bye, Rolf