The other thing I noted was in
aio.c
aioFini() did not check for errno == EINTR on close(), perhaps we don't care.
n= select(maxFd, &rd, &wr, &ex, &tv); we don't check for EINTR, perhaps we don't care. I'll note on the select() if we choose that logic that EINTR is a fairly usual case. Mind I wonder if we have to do the select again to fully service any pending interrupts?
and nanosleep(&rqtp, &rmtp); /* EINTR here, but likely we want to wake up? */
On Apr 2, 2007, at 3:57 PM, Andreas Raab wrote:
aioSleep
-- ======================================================================== === John M. McIntosh johnmci@smalltalkconsulting.com Corporate Smalltalk Consulting Ltd. http://www.smalltalkconsulting.com ======================================================================== ===