Re: [gamin] config parsing changes committed

[John McCutchan <ttb tentacle dhs org>]

On Fri, 2005-08-05 at 14:41 -0400, Daniel Veillard wrote:
> On Fri, Aug 05, 2005 at 02:28:50PM -0400, John McCutchan wrote:
> > 	2) Else, use the default gamin poll timeout which is 0 seconds.
> 
>   That's something you said already and to me is half wrong.
> There is 2 timouts in the general model:
>    - the one after which you deliver the event to the application (t)
>    - in polling mode the time beween 2 checks (T)
> 
>  in the gam_poll mode:
>    - t = 0, T = 1 second
>  in dnotify mode:
>    - t = 0, T is not used
>  in inotify old code
>    - t = 0, T is not used
>  in inotify new code code
>    - t = 100ms, T is not used
> 
>   To me the poll timeout, which is how often the ressource is polled is not
> 0 it is one second. Saying it's 0 second is misleading, everybody would expect
> that to be continuous polling.

This is the code,

/* If not enough time has passed since the last time we polled this
node, stop here */
    if (node->lasttime && (current_time - node->lasttime) <
        node->poll_time)
        return 0;

Maybe timeout is the wrong word. I guess it's more of a poll limiter. I
had originally set the poll limiter to 1 seconds, but that caused the
test suite to fail. So I had to set it to 0. What the option does is
limit how often a poll can happen. So, in that sense, we do want the
default to be 0.

Also, based on how it is implemented, I don't see how we can have
sub-second limiting.

-- 
John McCutchan <ttb tentacle dhs org>