[Xorp-users] select timeout question

[Ben Greear]

Pavlin Radoslavov wrote:
> Ben Greear <greearb at candelatech.com> wrote:
> 
>> If you pass a timeout of 0 to selector::wait_and_dispatch, it causes select to wait 'forever'.
>> Is that on purpose or a bug?  If you want to wait for zero time in select, pass a time-val
>> with all zero values (not 0, which is actually treated as NULL).
>>
>>      if (timeout == 0 || *timeout == TimeVal::MAXIMUM()) {
>> 	n = ::select(_maxfd + 1,
>> 		     &testfds[SEL_RD_IDX],
>> 		     &testfds[SEL_WR_IDX],
>> 		     &testfds[SEL_EX_IDX],
>> 		     0);
>>      } else {
> 
> I believe it is on purpose.
> Part of the logic behind it is probably because select(2) with
> "timeout" argument of NULL also blocks forever.
> One extreme scenario that comes to mind is, say, if there are no
> timeout events scheduled. In that case you want to select() until a
> file descriptor is ready (for reading/writing/etc).
> 
> Strictly speaking, the "timeout" argument doesn't need to be a
> pointer; instead, it could be pass-by-reference argument.
> In fact, the more I think, the more I believe it should be
> pass-by-reference. Then, if some code really wants to select(2)
> without timeout, the caller should use TimeVal::MAXIMUM() as an
> argument.
> 
> And yes, TimeVal::ZERO() is used typically to schedule a timer that
> will expire immediately.

Bleh, I see.  Timeout is a pointer.  Would have been more
clear if the code was:  timeout == NULL

Ben

> 
> Thanks,
> Pavlin
> 
>> -- 
>> Ben Greear <greearb at candelatech.com>
>> Candela Technologies Inc  http://www.candelatech.com
>>
>> _______________________________________________
>> Xorp-users mailing list
>> Xorp-users at xorp.org
>> http://mailman.ICSI.Berkeley.EDU/mailman/listinfo/xorp-users


-- 
Ben Greear <greearb at candelatech.com>
Candela Technologies Inc  http://www.candelatech.com