Re: [PATCH.HURD] Limit the minimum timeout delay to 1us for hurdselect.c to avoid problems
On Thu, Nov 01, 2012 at 06:20:50PM +0100, Svante Signell wrote:
> OK, 1ms seems to be a little long. If the time-resolution had been 1us
> it could have simplified the problem a lot.
This is what my select_timeout branches in some hurd repositories is all
about BTW. But with the recent work on pthreads, which showed other
problems caused by the way select is implemented, I intend to rewrite it
(the big problem is the amount of dead name notifications received by
servers when select is very often used, as is the case in completely
event driven applications).
There is no simple solution to this problem, it requires a lot more
efforts than these ugly hacks.